Deploying the BIG-IP Dual-Stack Data Center Firewall With F5

Deploying the BIG-IP Dual-Stack Data Center Firewall With F5
F5 Deployment Guide
Deploying the BIG-IP Dual-Stack Data Center Firewall With F5
Advanced Firewall Manager
Welcome to the F5 BIG-IP data center firewall Deployment Guide. This document provides guidance on configuring BIG-IP with AFM
(Advanced Firewall Manager) and LTM (Local Traffic Manager) as a high-security, high-availability, high-performance dual-stack data
center network firewall and IPv6/IPv4 gateway.
BIG-IP AFM is an ICSA Labs-certified network firewall which provides advanced network-protection capabilities meeting the needs of
all organizations, including those with PCIDSS, FIPS, and HIPAA compliance requirements.
BIG-IP AFM and LTM provide superior security and functionality for organizations integrating IPv6 into their network architecture and
operations.
This Deployment Guide includes extensive design information to help you bring BIG-IP security and performance to your existing
networks.
For more information on the F5 BIG-IP platform, see http://www.F5.com/products/big-ip.
For more information on the F5 BIG-IP Advanced Firewall Manager (AFM), see
https://F5.com/products/modules/advanced-firewall-manager.
Products and Versions
Product
BIG-IP system (LTM, AFM)
BIG-IP iApp template
Deployment Guide version
Last updated
Version
11.6 - 12.0
f5.datacenter_firewall_dg_quick_start.v1.0.0rc1
1.2 (see Document Revision History on page 74)
02-05-2016
Important: Make sure you are using the most recent version of this deployment guide, available at
http://f5.com/pdf/deployment-guides/f5-data-center-firewall-dg.pdf.
To provide feedback on this Deployment Guide or other F5 solution documents, contact us at [email protected]
Contents
The BIG-IP Application Delivery Firewall
3
Prerequisites and configuration notes
3
Deployment Guide Model Network Architecture
5
Allocating IP Addresses
6
Performing the initial configuration of the BIG-IP system for AFM
Configuring a DHCPv6 Relay
7
10
Performing the Basic AFM Configuration for Edge Firewall Use
11
Creating the Network Firewall Address Lists
11
Creating the Network Firewall Port Lists
13
Creating the Network Firewall Rule list
14
Creating the ICMP Rule list
15
Creating the Network Firewall Policies
18
IP Intelligence 20
Configuring AFM High-Speed Logging
23
Preparing for outbound traffic
24
Activating BIG-IP AFM Firewall Mode
26
Applying Fundamental Firewall Protection
27
Securing Outbound Traffic
28
Securing an Email Gateway
30
Securing a Multi-Tier Web Application with Presentation Servers in a DMZ
32
Configuring DMZ1 Administrative and Backend Traffic
33
Creating a Protocol Security Profile
37
Configuring the BIG-IP system to send web application traffic into the DMZ
37
Additional information: Admitting Application Traffic to a DMZ
41
Securing a Branch-Office, Cloud-Infrastructure, or Business-Partner Link
Additional Information: Securing Links to Remote Networks
42
45
Defending the BIG-IP system Itself with AFM
46
Protecting the BIG-IP Management Port with AFM
46
Deep Packet Inspection with AFM (Blocking Teredo)
48
Appendix A: Supplemental Information
49
Dual-Stack IPv6/IPv4 Networking 49
BIG-IP AFM Overview
50
Network Firewall
51
IP Intelligence
56
Denial-of-Service (and DNS) Protection
59
Troubleshooting AFM
60
Appendix B: Storing IP Intelligence Address Lists in BIG-IP Data Groups
62
Appendix C: Securing AFM Domain Name Resolution
66
Configuring a BIG-IP DNS Cache Validating Resolver
66
Glossary of Terms
71
Document Revision History
74
F5 Deployment Guide
2
Data Center Firewall
The BIG-IP Application Delivery Firewall
For years, sophisticated organizations have relied upon the F5 BIG-IP product family to manage and deliver application network traffic
reliably, securely, and quickly. Very often this has included managing and balancing network traffic to and through network firewall
devices, many of which cannot approach the performance of the BIG-IP system itself. Now the BIG-IP Advanced Firewall Manager
(AFM) module provides a comprehensive, ICSA Labs-certified network firewall solution to protect the data center. BIG-IP AFM is
integrated with BIG-IP Local Traffic Manager (LTM) and other BIG-IP modules.
During the long incubation period of IPv6 the F5 BIG-IP has gained recognition as the most effective tool for building transparent,
secure, high-performance dual-stack corporate networks. The BIG-IP system’s IPv6 fluency simplifies network security planning,
operations, and management.
The F5 BIG-IP constitutes an Application Delivery Firewall platform which provides a unified view of Layers 3 through 7+ for both
general and ICSA-mandatory reporting and alerting as well as integration with SIEM systems. The BIG-IP system’s full-proxy
architecture negates so called “advanced evasion technique” attacks which bypass many common firewalls.
The BIG-IP AFM module operates chiefly at OSI Layers 2 through 4, with significant functions at Layers 5 to 7. Additional BIG-IP
modules—the Application Security Manager (ASM) and Access Policy Manager (APM)—provide application-layer firewall and AAA
gateway services at Layers 7+ as well as TLS VPN services (L3+).
Combining the BIG-IP APM with AFM and/or ASM is particularly valuable, because APM enables identity-based security. With APM
you may tune policy for authorized users rather than subject everyone to least-common-denominator firewall rules. More critically, you
may track security alerts and issues back to individuals in many cases. You may also resolve advanced-persistent-threat, credential or
identity theft, and even malfeasance problems much more effectively when you associate user identity to network activity.)
Because L2–7 network firewall security is foundational for information system protection, this Deployment Guide shows how to
implement key policies to protect a typical data center network architecture using BIG-IP with AFM and LTM. You may replicate and
adjust these policies to protect your own network and the information systems using it.
Prerequisites and configuration notes
The following are general prerequisites, assumptions, and notes about the configuration described in this guide.
hh F
or the maximum benefit from this Deployment Guide, you should be familiar with network security concepts and the
basics of the BIG-IP platform.
he material in this Guide is not a substitute for the product documentation available at:
T
https://support.F5.com/kb/en-us/products/big-ip-afm/versions.11-6-0.html.
hh Y
ou must have at least one data center with an Internet link. The configuration in this guide supports public-facing and
internal applications.
hh W
e assume your users access applications in the data center as well as external services via the data center’s Internet link.
You may also have VPN links to branch-office, cloud-infrastructure, or business-partner networks with independent Internet
connectivity.
hh T
his guide describes an implementation in which the F5 BIG-IP with AFM and LTM is deployed between your primary
ISP link and your intranet to secure inbound and outbound traffic against intrusions and DoS attacks (the BIG-IP system
supports single and multiple ISP links with or without BGP routing). The BIG-IP AFM constitutes the primary edge firewall.
You may also use AFM as an interior firewall.
hh T
his guide shows you how to manage and secure both IPv6 and IPv4 traffic simultaneously. Both types of traffic are active
on nearly all LANs today.
hh N
etwork security deployments implement specific rules and practices to mitigate the challenges of a threat model. For
purposes of this Deployment Guide we do not explore the subject very deeply. Rather, we draw guidance from the F5 DDoS
Protection Reference Architecture along with NIST Special Publications 800 41 rev1—Guidelines on Firewalls and Firewall
Policy and 800-119—Guidelines for the Secure Deployment of IPv6, and the Payment Card Industry (PCI) Data Security
Standard Version 3.0 Requirements. We also used the BIG-IP Systems Network Firewall Guide for ICSA Certification and
other F5 reference and guidance documents, in addition to RFC7123, RFC4890, and other sources.
F5 Deployment Guide
3
Data Center Firewall
hh D
ata center firewalls support both North-South (NS) and East-West (EW) use cases. North-South refers generally to network
traffic between the Internet and the organizational intranet. East-West refers generally to network traffic between portions of
the intranet which constitute different trust environments. Trust boundaries are defined by security policy (much of which is
expressed in firewall rules) and need not mirror physical network layout.
hh T
he BIG-IP AFM includes a number of additional features that are not a part of this guide, such as SNMP trap configuration,
connection-eviction policy, SIP Protocol and DoS protection, and IPFIX setup.
hh You must have access to the BIG-IP web-based Configuration utility (GUI) and command line.
hh T
here is an iApp template available to make much of the initial configuration easier. See
https://devcentral.f5.com/codeshare/data-center-firewall-quick-start-iapp-template. The iApp is not required, but will save
configuration time.
This deployment guide covers the following topics:
• Performing the initial configuration of the BIG-IP system for AFM on page 7,
• Performing the Basic AFM Configuration for Edge Firewall Use on page 11,
• Securing an Email Gateway on page 30,
• Securing a Multi-Tier Web Application with Presentation Servers in a DMZ on page 32,
• Securing a Branch-Office, Cloud-Infrastructure, or Business-Partner Link on page 42,
• Defending the BIG-IP system Itself with AFM on page 46,
• Deep Packet Inspection with AFM (Blocking Teredo) on page 48
For more detail on the concepts and configuration presented in this guide, see Appendix A: Supplemental Information on page 49.
F5 Deployment Guide
4
Data Center Firewall
Deployment Guide Model Network Architecture
Figure 1 shows key elements of the model network architecture which is the basis for the examples in this Deployment Guide.
In this diagram, the BIG-IP system is the gateway to the Internet for the data center and much of the intranet so the default routes
(IPv6 and IPv4) on most devices point toward it (mostly via intermediate routers/L3 switches). However, branch office networks may
have independent Internet access.
Network-management (e.g., SNMP) tools, system administrators’ workstations, and infrastructure devices’ management ports are
homed on management network subnets.
The model network supports split-horizon DNS (separate Internet and intranet views) though we do not specify implementation
details. IPv6 devices learn DNS server addresses from DHCPv6. A spam-filtering mail server handles both incoming and outgoing
SMTP traffic.
Third-party security audits review static configuration files so this guide shows how to enforce explicit firewall policies on BIG-IP
listeners. You will discourage sloppy configuration practices by running AFM in Firewall mode. In Firewall mode, no virtual server or
self IP object will process any traffic until you enforce a named firewall policy on it (having at least one rule with an action of “Accept”).
In the configuration described in this guide, the example organization is forbidden to do business with entities in certain countries
listed by the US State Department under the International Traffic in Arms Regulations (ITAR regime) codified at 22 CFR §126.1. The
firewall policy must exclude traffic from those countries.
Finally, there is a log server which will accept logs as fast as the BIG-IP AFM can send them.
Figure 1: Data Center Firewall Deployment Guide Model Network Architecture
F5 Deployment Guide
5
Data Center Firewall
Allocating IP Addresses
With IPv6 as with IPv4 you will have two main blocks of addresses: public addresses reachable from the Internet and private
addresses used in your intranet. All traffic between the Internet and your intranet requires address translation; this greatly aids security
filtering and minimizes the cost of switching ISPs. Public addresses of outside (Internet) network correspondents will be visible in
the intranet as source-address on inbound traffic and destination-address on outbound traffic. This averts any hassle with DNS and
makes your logs easy to search and analyze.
Commonly, your ISP will give you a PA (Provider Assigned) IPv6 public address space and you will use a ULA (Unique Local Address)
IPv6 private address space in your intranet. If you have PI (Provider-Independent) IPv6 address space, you may subnet it into public
and private blocks or use it just for public addresses.
The examples in this document illustrate one way to align IPv4 and IPv6 private subnet numbers for the convenience of network
administrators. For instance, one of the data center VLANs carries the paired subnets fdf5:f:5:c001::/64 and 10.1.1.0/24. The same
digits are used for the third octet of the IPv4 subnet number and the last three columns of the fourth chunk of the IPv6 subnet number.
The first column in the fourth chunk is a mnemonic code for a region of the intranet (‘c’ for data center—elsewhere we use ‘d’ for the
DMZ); the intranet region also selects the first two octets of the IPv4 subnet number (10.1 for data-center—compare 172.30 for DMZ).
While the explanation of this scheme may seem complicated, it is easier to put into practice.
Much network firewall policy is concerned with packet addressing. When you adapt examples from this document to create your
own AFM configuration, you may find the following table useful. Note that you won’t replace addresses in lists of standard subnets like
island-nets (defined later in this document).
Description
Deployment Guide Example
Public IPv4 subnet
192.0.2.0/24
Public IPv6 prefix + subnet
2001:db8:16d:2::/64
Private IPv4 interior subnets
10.W.X.0/24
Private IPv4 “management network” block of subnets
192.168.192.0/18
Private IPv4 DMZ subnets
172.30.X.0/24
Private IPv6 prefix + subnets
fdf5:f:5:ZZZZ::/64
Private IPv6 “management network” block of subnets
fdf5:f:5:f000/54
BIG-IP management subnet
192.168.193.0/24
BIG-IP HA subnet
192.168.245.0/24
F5 Deployment Guide
6
Your value
Data Center Firewall
Performing the initial configuration of the BIG-IP system for AFM
In this section, we provide guidance on the initial setup of the BIG-IP system for the Advanced Firewall Manager module.
i
Important In the following section, and throughout the guide, we use example values and object names based on our
configuration. Change these to the appropriate values and names for your configuration.
1. Install your BIG-IP devices according to F5 documentation
It is outside the scope of this document to provide instructions on installing the BIG-IP system in your network. Refer to the
BIG-IP documentation available on the F5 technical support web site: https://support.f5.com.
2. U
se a redundant pair of BIG-IP devices (recommended)
This guide is written with the assumption you have a pair of BIG-IP devices for redundancy. In our examples, our BIG-IP devices
are named big-s1 and big-s2. F5 ScaleN deployments with additional devices are also feasible, but not detailed in this guide.
3. Configure
the management addresses
Each BIG-IP must have an IPv4 or IPv6 management-port address and management default route. Typically you configured
these addresses during the initial BIG-IP configuration. You can find the settings in System > Platform. In the User
Administration area, leave SSH Access set to Enabled and SSH IP Allow set to * All Addresses. We show how to secure
management SSH access later. The following table shows our example devices and their associated management IP addresses.
BIG-IP device
Management IP address (default route)
big-s1 (big-s1.example.net)
192.168.193.21/24 (192.168.193.1)
big-s2 (big-s1.example.net)
192.168.193.22/24 (192.168.193.1)
4. E
nsure AFM is provisioned on your BIG-IP system
Go to System > Resource Provisioning > Configuration, and make sure Advanced Firewall (AFM) is set to Nominal.
If it is not, select Nominal from the list, click Submit and then wait for the BIG-IP system to update.
5. C
onfigure VLANs on your BIG-IP system
On each BIG-IP device, configure the following VLANs. For VLAN configuration, go to Network > VLANs.
VLAN
Remarks
external
link to Internet gateway [standard]
internal
link to intranet/data-center gateway (router/L3-switch) [standard]
HA
links redundant BIG-IP devices [standard]. You may use a (link aggregation) trunk to downmux BIG-IP HA (ConfigSync
and Mirroring) traffic across multiple L2 ports for added performance. Consult BIG-IP TMOS Concepts for details.
6. C
onfigure the BIG-IP self IP addresses on each device
On the first BIG-IP device (big-s1 in our example) use the following table for guidance on creating self IP addresses, using the
addresses appropriate for your implementation. All self IP addresses should be created in the /Common partition.
For Self IP configuration, go to Network > Self IPs.
Name
IP Address
Netmask
VLAN
Port Lockdown
Traffic Group
self-ext-s1-4
192.0.2.3
255.255.255.0
external
Allow Default
traffic-group-local-only
self-ext-s1-6
2001:db8:16d:2::3
ffff:ffff:ffff:ffff::
external
Allow Default
traffic-group-local-only
self-int-s1-4
10.1.245.2
255.255.255.0
internal
Allow Default
traffic-group-local-only
self-int-s1-6
fdf5:f:5:c245::2
ffff:ffff:ffff:ffff::
internal
Allow Default
traffic-group-local-only
self-ha-s1
192.168.245.21
255.255.255.0
HA
Allow Default
traffic-group-local-only
On the second BIG-IP device (big-s2 in our example) create self IP addresses using the following table as guidance. All self IP
addresses should be created in the /Common partition.
Name
IP Address
Netmask
VLAN
Port Lockdown
Traffic Group
self-ext-s2-4
192.0.2.4
255.255.255.0
external
Allow Default
traffic-group-local-only
self-ext-s2-6
2001:db8:16d:2::4
ffff:ffff:ffff:ffff::
external
Allow Default
traffic-group-local-only
self-int-s2-4
10.1.245.3
255.255.255.0
internal
Allow Default
traffic-group-local-only
self-int-s2-6
fdf5:f:5:c245::3
ffff:ffff:ffff:ffff::
internal
Allow Default
traffic-group-local-only
self-ha-s2
192.168.245.22
255.255.255.0
HA
Allow Default
traffic-group-local-only
F5 Deployment Guide
7
Data Center Firewall
On both devices create floating self IP addresses using the following table as guidance.
All self IP addresses should be created in the /Common partition.
Name
IP Address
Netmask
VLAN
Port Lockdown
Traffic Group
self-ext-float-4
192.0.2.2
255.255.255.0
external
Allow Default
traffic-group-1
self-ext-float-6
2001:db8:16d:2::2
ffff:ffff:ffff:ffff::
external
Allow Default
traffic-group-1
self-int-float-4
10.1.245.1
255.255.255.0
internal
Allow Default
traffic-group-1
self-int-float-6
fdf5:f:5:c245::1
ffff:ffff:ffff:ffff::
internal
Allow Default
traffic-group-1
7. C
onfigure NTP, DNS, and named.conf
On each device: configure NTP (Network Time Protocol) (SOL3122). Also configure DNS Lookup Server List plus BIND
Forwarder Server List (SOL13205). Then correct each BIG-IP’s named.conf file per SOL12224. Optionally configure a remote
syslog server in System / Logs / Configuration / Remote Logging.
8. U
pdate the IP Geolocation database
Download the latest IP Geolocation database update per SOL11176. Install it on each device.
9. C
onfigure High Availability
You can add both BIG-IP devices to a high availability Sync-Failover group using the Configuration utility on one of them.
The BIG-IP HA link in our example crosses only one VLAN so we just use IPv4 for HA communications. Use of IPv6 on BIG-IP
HA links is supported but slightly less efficient. If you decide to use IPv6 for BIG-IP high availability traffic, review SOL15816.
Log in to the first BIG-IP device, big-s1 in our example.
a. On the Main tab, click Device Management > Devices.
• Click big-s1.example.net (Self).
• On the Menu bar, click Device Connectivity > Config Sync.
• From the Local Address list, select 192.168.245.21 (HA).
• Click Update.
b. On the Menu bar, click Device Connectivity > Failover Network.
• Ensure the Failover Unicast List contains 192.168.193.21 (VLAN Management Address) and 192.168.245.21
(VLAN HA). If these addresses do not appear in the list, click the Add button to add the missing address(es).
• Leave the Use Failover Multicast Address unchecked (do NOT check this box).
• Click Update.
c. On the Menu bar, click Device Connectivity > Mirroring.
• From the Primary Local Mirror Address list, select 192.168.245.21 (HA).
• From the Secondary Local Mirror Address list, select 10.1.245.2 (internal).
• Click Update.
d. On the Main tab, click Device Management > Device Trust > Peer List.
• Click Add.
• In the Device IP Address field, type 192.168.99.22 (the management IP address of big-s2).
• In the Administrator Username field, type admin.
• In the Administrator Password field, type the associated password.
• Click Retrieve Device Information. The other BIG-IP device's host name (big-s2.example.net in our example)
should appear in the Device Properties Name field.
• Click Finished.
e. On the Main tab, click Device Management > Device Groups.
• Click Create.
F5 Deployment Guide
8
Data Center Firewall
• In the Name field, type ha-group-1.
• From the Group Type list, select Sync-Failover.
• In the Members area, use the Add button (<<) to move both big-s1.example.net and big-s2.example.net from
Available to Includes.
• Check the Network Failover box.
• Leave Automatic Sync unchecked (NOT checked).
• Click Finished.
f.
On the Main tab, click Device Management > Overview.
• In the Device Groups area, click ha-group-1.
• In the Devices area, click big-s1.example.net (Self).
• In the Sync Options area that appears, click Sync Device to Group.
• Check the Overwrite Configuration box.
• Click Sync.
This completes the high availability configuration.
F5 DevCentral offers a good tutorial the BIG-IP HA setup: ScaleN: A Network Architect-Engineer’s Unofficial Guide To ScaleN
Clustering.
10. A
dd the IP routes to the BIG-IP system
On each BIG-IP device, add the following IP routes in the /Common partition. Note that if you encounter an error message
similar to 01070712:3: Cannot create static route: ::/0 gw 2001:db8:16d:2::1 on interface '' in rd0 - netlink error: 113 (No route
to host) when using TMOS 11.6, upgrade to 11.6 Hotfix 4 or later.
Name
Destination
Netmask
Resource
Gateway
default-4
0.0.0.0
0.0.0.0
Use Gateway…
192.0.2.1
default-6
::
::
Use Gateway…
2001:db8:16d:2::1
intranet-4-pA
10.0.0.0
255.0.0.0
Use Gateway…
10.1.245.7
intranet-4-pB
172.16.0.0
255.240.0.0
Use Gateway…
10.1.245.7
intranet-4-pC
192.168.0.0
255.255.0.0
Use Gateway…
10.1.245.7
intranet-6
fdf5:f:5::
ffff:ffff:ffff::
Use Gateway…
fdf5:f:5:c245::7
11. O
n each BIG-IP device, use the virtual console or an SSH client such as PuTTY to access the BIG-IP command line and
execute the following TMSH commands to enable IPv6 Neighbor Discovery on the intranet VLAN only. For more details see
SOL13580: Configuring neighbor discovery for IPv6. Leave the ‘A’ flag set in RA’s.
In the model network architecture, servers learn DNS settings from DHCPv6:
create /net router-advertisement ra-internal vlan internal enabled prefixes add { /Common/pfx-internal { prefix
fdf5:f:5:c245:: prefix-length 64 } router } managed other-config
If you set up a network without DHCPv6, substitute this command:
create /net router-advertisement ra-internal vlan internal enabled prefixes add { /Common/pfx-internal { prefix
fdf5:f:5:c245:: prefix-length 64 } router }
Save your changes:
save /sys config
F5 Deployment Guide
9
Data Center Firewall
Configuring a DHCPv6 Relay
IPv6 emphasizes dynamic configuration. Still, as with IPv4, devices using IPv6 may be given static IP addresses1, default routes, and/
or DNS settings. However, in many intranets even devices with static IPv6 addresses use DHCPv6 to get DNS settings. When you
operate DHCPv6 servers you should make the BIG-IP device a DHCPv6 relay. If necessary, you can also configure a IPv4 DHCP relay.
In the model network, the DHCPv6 server address is fdf5:f:5:c006::9. Additional DHCPv6/DHCP servers are supported; you can
simply add them to the pool(s). Use the following table to configure the appropriate objects. For specific instructions, see the Help
tab or the product documentation.
hile your IP addresses will almost always be different from our examples, you may find it easier to follow the
 Tip: W
configuration in this guide if you use the same object names as in our examples. Some objects are called from other
objects, which in turn are called from another object (and so on). Our guidance always refers to our example names.
Pools (Navigate to Local Traffic > Pools)
Name
DHCPv6-pool
Health Monitor
Select gateway_icmp.
New Members
Node Name
dhcpv6-server1
Node Address
fdf5:f:5:c006::9
Service Port
547
Virtual Servers (Navigate to Local Traffic > Virtual Servers)
Name
DHCPv6-relay-vs
Type
DHCP
Destination Address
ff02::1:2 (IPv6 Default)
DHCP Profile
From the first list, select DHCPv6. From the second list, under /Common select dhcpv6.
VLAN and Tunnel Traffic
Select Enabled On, and then select only the Internal VLAN you created.
Default Pool
Select the DHCPv6 pool you created
Note: You may download an iApp to perform much of the following configuration from:
https://devcentral.f5.com/codeshare/data-center-firewall-quick-start-iapp-template
Use the virtual console or an SSH client such as PuTTY to access the BIG-IP command line on each device and execute the following
TMSH commands:
modify /sys syslog iso-date enabled
modify /sys db tm.icsastricttcpforwarding value enable
modify /sys db dos.dropv4mapped value true
modify /ltm global-settings general snat-packet-forward enabled
save /sys config
From this point, unless otherwise noted you may add configuration changes to one BIG-IP device and then propagate them using
configsync.
1
It is customary for servers in LTM load balancing pools (members, nodes) to have static IP addresses, though LTM will let you identify nodes and
members by FQDN mapped to IP address by DNS
F5 Deployment Guide
10
Data Center Firewall
Performing the Basic AFM Configuration for Edge Firewall Use
The first job of a data center firewall is to block everything coming in until you decide what to allow. Even after that, the firewall
must block unwanted traffic trying to get in through any ports you open. The firewall's second job is to let traffic go out, except for
suspicious traffic trying to reach destinations you don't approve of. Think of your data center firewall as a fortress with eagle-eyed
sentries. You can get out; you can let your friends in; but it frustrates your enemies and their spies.
With BIG-IP AFM, you layer application-specific security policy onto a base of a global policy aimed at generic threats. This section
shows how to establish that base layer.
For a review of AFM structure and configuration, see BIG-IP AFM Overview on page 50 and subsequent topics.
ou can download an iApp to perform much of the following configuration. See
 Note: Y
https://devcentral.f5.com/codeshare/data-center-firewall-quick-start-iapp-template
Creating the Network Firewall Address Lists
Reusable elements make firewall rules and policies more efficient and maintainable. For instance, address lists let you replace blocks
of numbers with mnemonic names and streamline updates to addresses used in multiple rules.
Use the following guidance to create all of the following address lists, adding subnets (CIDR notation) individually. Instead of clicking
Finished after adding each list, you can click Repeat to save time.
Address Lists (Navigate to Security > Network Firewall > Address Lists)
Name
island-nets
Description
Standard RFC6890
Notes (not a part of the configuration)
Addresses/Regions
Type or copy and paste the following addresses into the Add new entry box:
100.64.0.0/10
::/96
2001:10::/28
127.0.0.0/8
100::/64
2001:20::/28
169.254.0.0/16
2001:2::/48
2001:db8::/32
May not be src or dst of packet on the wire.
(::/96 catches IPv6 self and loopback.)
::ffff:0:0/96 is purposefully omitted until allowed
by a future F5 TMOS update1.
240.0.0.0/4
Name
unicast-nets
Description
Standard
Addresses/Regions
0.0.0.0/1
192.0.0.0/3
128.0.0.0/2
2000::/3
Name
mcast-nets
Description
Standard
Addresses/Regions
224.0.0.0/4
Name
nonpub-nets
Description
Standard RFC6890
Addresses/Regions
(We “notch out” some IPv4 nets.)
Always invalid as src address
ff00::/8
0.0.0.0/8
192.0.2.0/24
203.0.113.0/24
10.0.0.0/8
192.168.0.0/16
64:ff9b::/96
172.16.0.0/12
198.18.0.0/15
fc00::/7
192.0.0.0/24
198.51.100.0/24
Name
bad-6to4-nets
Description
Standard
Addresses/Regions
As of 2015, covers all public IPv6 unicast.
2002::/24
2002:ac10::/28
2002:c612::/31
2002:0a00::/24
2002:c000::/40
2002:c633:6400::/40
2002:6440::/26
2002:c000:0200::/40
2002:cb00:7100::/40
2002:7f00::/24
2002:c0a8::/32
2002:e000::/19
Not allowed from or to public Internet.
We treat TEST-NET-n as private-use.
2002:a9fe::/32
1
Refer to f5 ID456376. This Guide sets TMOS db variable dos.dropv4mapped to avert any security issue.
F5 Deployment Guide
11
Data Center Firewall
Address Lists (Navigate to Security > Network Firewall > Address Lists)
Name
intranet-nets
Notes (not a part of the configuration)
Description
Example
Addresses/Regions
These addresses are our examples. Adjust for your network if necessary.
10.0.0.0/8
192.168.0.0/16
172.16.0.0/12
fdf5:f:5::/48
Name
net-mgmt-nets
Description
Example
Addresses/Regions
These addresses are our examples. Adjust for your network if necessary.
192.168.192.0/18
all-dc-nets
Description
Example
Addresses/Regions
These addresses are our examples. Adjust for your network if necessary.
our-pub-nets
Description
Example
Addresses/Regions
These addresses are our examples. Adjust for your network if necessary.
outbound-snat-nets
Description
Example
Addresses/Regions
These addresses are our examples. Adjust for your network if necessary.
mailserver
Description
Example
Addresses/Regions
These addresses are our examples. Adjust for your network if necessary.
To see “mailserver” in AFM rules rather than the
addresses
fdf5:f:5:c007::25/128
Name
ITAR-countries
Description
From 22 CFR 126.1 updated 2015–03–27
Addresses/Regions
Belarus (BY)
F5 Deployment Guide
Addresses used to SNAT outbound connections
from intranet to Internet
2001:db8:16d:2::240/126
Name
10.1.7.25/32
Public Internet-accessible subnets
2001:db8:16d:2::/64
Name
192.0.2.240/30
All Data Center subnets
fdf5:f:5:c000::/54
Name
192.0.2.0/24
Device management, SNMP tools, RADIUS
servers, etc.
fdf5:f:5:f000::/54
Name
10.1.0.0/16
All intranet subnets (includes all-dmz-nets;
specific rules control DMZ access)
Korea, Democratic People's
Republic of (KP)
Congo, The Democratic
Lebanon (LB)
Republic of the (CD)
Liberia (LR)
Cote D'Ivoire (CI)
Libyan Arab Jamahiriya (LY)
Cuba (CU)
Somalia (SO)
Eritrea (ER)
Sudan (SD)
Iran, Islamic Republic of (IR)
Syrian Arab Republic (SY)
Iraq (IQ)
Venezuela (VE)
12
This is an example of how you can use
geolocation, and is optional.
When you really implement ITAR restrictions,
update this list and the date in the Description.
Data Center Firewall
Creating the Network Firewall Port Lists
Use the following table for guidance to create all of the following port lists, adding port numbers individually. Again, because there are
so many lists, instead of clicking Finished after adding each list, you can click Repeat to save time.
Port Lists (Navigate to Security > Network Firewall > Port Lists)
Notes (not a part of the configuration)
Name
Port
dns
53
ftp
21
http
80
https
443
imap-tls
993
kerberos
88, 464
ldap
389
ldaps
636
ms-ad-gc
3268, 3269
ms-sql
1433
ntp
123
smtp-mta
25
smtp-mua
587
smtp-all
25, 587
snmp-mgr
161, 10161
Only network managers may send to these ports.
smtp-trap
162, 10162
10162 for DTLS/TLS
ssh
22
syslog-udp
514
syslog-tls
6514
tripwire
1169
udp-dhcp
67, 68, 546, 547
udp-EW-basic
dns, ntp, snmp-trap, syslog-udp,
tripwire
FTP uses additional ports—ALG required
Active Directory Global Catalog (LDAP on the wire)
Once you have created a Port list, you can use that list in a new list. Begin typing and
the Port list will appear in the box, preceded by /Common.
DMZ servers typically access self-secured critical intranet services on these ports. If you
do not use Tripwire you may omit the port.
tcp-EW-basic
dns, smtp-mua, syslog-tls, tripwire
If you do not use Tripwire you may omit the port. DMZ servers typically access selfsecured critical intranet services on these ports.
udp-EW-generic
kerberos
Kerberos is often needed with tcp-EW-generic and tcp-rdp-smb.
basic+generic+rdp+smb is a good starting point for screening branch-office or
workstation nets.
tcp-EW-generic
http, https, imap-tls, kerberos, ldap,
ldaps, ms-ad-gc, smtp-mta, ssh
When DMZ servers must access specific services (e.g., backend resources) on these
ports you should create corresponding BIG-IP virtual servers, or at least add specific
firewall rules (e.g., src dmz:any dst w.x.y.z:https). However, when that is too difficult, you
might permit basic+generic and rely on (destination) host/service security.
tcp-rdp-smb
445, 3389
Remote desktop, file server access, and so on.
u-t-tsp-ayiya
3653, 5072
TSP or AYIYA tunnels (VPN/policy evasion)
This completes the Network Firewall Port list configuration.
Creating the Network Firewall Rule
The next task is to create the following iRule. This iRule is tied to the global IP Intelligence policy you define later, but you create it now
so you can reference it in network firewall rules.
F5 Deployment Guide
13
Data Center Firewall
From the Main tab, expand Local Traffic and then click iRules. Click the Create button. Give the iRule a unique name (we use
afm-no-evil-dst), and then use the following code in the Definition section, omitting the line numbers:
1
2
3
4
5
6
7
8
9
10
11
12
13
when FLOW_INIT {
set okay [list "cloud_provider_networks"]
set r [IP::intelligence [IP::local_addr]]
if {[lsearch -exact $r "_WHITELIST_"] >= 0} { return }
foreach {cat} $r {
if {[lsearch -exact $okay $cat] < 0} {
#log local0.info "denied outbound to [IP::local_addr] = ${cat}"
ACL::action reset ; #reject
return
}
}
}
Creating the Network Firewall Rule list
The next task is to create the Firewall rule lists. You first create the Rule list with a name and optional description, and then add rules
to it. As you edit each rule, extra fields appear when you change a list from Any to Specify.
For the first list, the following table details the configuration. For the subsequent rule lists, use the second table of values for guidance.
Rule Lists (Navigate to Security > Network Firewall > Rule Lists)
Click Finished and then from the Rule Lists table, click the Rule List you just created.
Rule List Name
global-forbid-nets
Rules
Click the Add button and add the following rules.
Rule List Name
control-from-Internet
F5 Deployment Guide
Name
no-src-island-mcast
Protocol
Any
Source > Address/Region
Add island-nets and mcast-nets
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
Any
Destination > Port
Any
Action
drop
Name
no-dst-island-4lnk
Protocol
Any
Source > Address/Region
Any
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
Add island-nets and 0.0.0.0/8
Destination > Port
Any
Action
drop
Rule Name
Protocol
Click Repeat and then add the following rule.
Click Finished
Source address
Src. Port
Src. VLAN
Dst. Address
Dst. Port
Action
Any
Accept
fe80::/10
Any
external
fe80::/10
ff02::/16
Any
fe80::/10
Any
external
Any
Any
Drop
no-src-nonpub-etc
Any
nonpub-nets
bad-6to4-nets
intranet-nets
Any
external
Any
Any
Drop
no-dst-nonpub-etc
Any
Any
Any
external
nonpub-nets
intranet-nets
Any
Drop
ok-mcast-link-global
Any
unicast-nets
Any
external
224.0.0.0/4
ff02::/16
ff0e::/16
Any
Accept
no-other-mcast
Any
Any
Any
external
mcast-nets
Any
Drop
ok-valid-src
Any
unicast-nets
Any
external
Any
Any
Accept
no-other-src
Any
Any
Any
external
Any
Any
Drop
ok-ipv6-link
Any
no-fwd-link
14
Data Center Firewall
Rule List Name
block-selfIP-mgmt
Rule Name
Protocol
Source address
Src. Port
Src. VLAN
Dst. Address
Dst. Port
Action
no-https
TCP
Any
Any
Any
Any
HTTPS
Drop
no-ssh
TCP
Any
Any
Any
Any
SSH
Drop
no-snmp-u
UDP
Any
Any
Any
Any
snmp-mgr
Drop
no-snmp-t
TCP
Any
Any
Any
Any
snmp-mgr
Drop
ok-dhcp
UDP
Any
udp-dhcp
Any
Any
udp-dhcp
Accept
no-dst-nonpub-etc
Any
Any
Any
Any
nonpub-nets
intranet-nets
192.88.99.0/24
Any
Reject
no-send-mcast
Any
Any
Any
Any
mcast-nets
Any
Reject
no-tsp-ayiya-t
TCP
Any
Any
Any
Any
u-t-tsp-ayiya
Reject
no-tsp-ayiya-u
UDP
Any
Any
Any
Any
u-t-tsp-ayiya
Reject
ok-smtp-mailserver
TCP
mailserver
Any
Any
Any
smtp-all
Accept
permit-dhcp
filter-outbound
For this Rule, you also need to attach the iRule you created in Creating the Network Firewall Rule on page 13
no-smtp-others
TCP
Any
Any
Any
Any
smtp-all
Reject
ok-tcp
TCP
intranet-nets
Any
Any
Any
Any
Accept
Any
Accept
Any
Reject
For this Rule, you also need to attach the iRule you created in Creating the Network Firewall Rule on page 13
ok-udp
UDP
intranet-nets
Any
Any
Any
For this Rule, you also need to attach the iRule you created in Creating the Network Firewall Rule on page 13
no-other
Any
Any
Any
Any
Any
Creating the ICMP Rule list
The next task is to create a global ICMP control rule list as described in the following table. For each rule in the table, after you select
the Protocol type you must choose ICMP type-and-code combinations and Add them to the message list.
Rule Lists (Navigate to Security > Network Firewall > Rule Lists)
global-icmp-control Rule list
Click Finished and then from the Rule Lists table, click the Rule List you just created.
Rule List Name
global-icmp-control
Rules
Click the Add button and add the following rules.
Name
no-ping6-snats
Protocol
ICMPv6 Make sure 58 appears in the box on the right.
ICMPv6 Message
Type
Code
Echo Request (128)
Source > Address/Region
any
Source > Port
any
Source > VLAN/Tunnel
any
Destination > Address/Region
Add outbound-snat-nets
Destination > Port
any
Action
drop
Name
ok-always-icmp6
Protocol
ICMPv6 Make sure 58 appears in the box on the right.
ICMPv6 Message
F5 Deployment Guide
any
Click Repeat and then add the following rule.
Type
Code (click Add after each)
Destination Unreachable (1)
Any
Packet Too Big (2)
Any
Time Exceeded (3)
Hop limit exceeded in transit (0)
Time Exceeded (3)
Fragment reassembly time exceeded (1)
Parameter Problem (4)
Erroneous header field encountered (0)
Parameter Problem (4)
Unrecognized Next Header type encountered (1)
Parameter Problem (4)
Unrecognized IPv6 option encountered (2)
Echo Request (128)
Any
Echo Reply (129)
Any
15
Data Center Firewall
Rule Lists (Navigate to Security > Network Firewall > Rule Lists)
Rules continued
Source > Address/Region
Any
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
Any
Destination > Port
Any
Action
Accept Decisively
Name
ok-local-icmp6
Protocol
ICMPv6 Make sure 58 appears in the box on the right.
ICMPv6 Message
Type
Code (then click Add)
Multicast Listener Query (130)
Any
Multicast Listener Report (131)
Any
Multicast Listener Done (132)
Any
Router Solicitation (133)
Any
Router Advertisement (134)
Any
Neighbor Solicitation (135)
Any
Neighbor Advertisement (136)
Any
Inverse Neighbor Discovery Solicitation Message (141)
Any
Inverse Neighbor Discovery Advertisement Message (142)
Any
Version 2 Multicast Listener Report (143)
Any
Certification Path Solicitation Message (148)
Any
Certification Path Advertisement Message (149)
Any
Multicast Router Advertisement (151)
Any
Multicast Router Solicitation (152)
Any
Multicast Router Termination (153)
Any
Source > Address/Region
fe80::/10 (If needed in your network, add destination prefixes like site-local multicast ff05::/16)
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
Add fe80::/10 and ff02::/16
Destination > Port
Any
Action
Accept Decisively
Name
no-unk-icmp6
Protocol
ICMPv6 Make sure 58 appears in the box on the right.
ICMPv6 Message
Click Repeat and then add the following rule.
Type
Code
Any
Any
Source > Address/Region
Any
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
Any
Destination > Port
Any
Action
Drop
Name
ok-snats-icmp4
Protocol
ICMP Make sure 1 appears in the box on the right. Note the change from ICMPv6 to ICMP.
ICMP Message
F5 Deployment Guide
Click Repeat and then add the following rule.
Click Repeat and then add the following rule.
Type
Code (then click Add)
Echo Reply (0)
No Code (0)
Destination Unreachable (3)
Net Unreachable (0)
Destination Unreachable (3)
Host Unreachable (1)
Destination Unreachable (3)
Fragmentation needed and Don’t Fragment… (4)
Destination Unreachable (3)
Communication Administratively Prohibited (13)
Time Exceeded (11)
Time To Live exceeded in transit (0)
Time Exceeded (11)
Fragment Reassembly Time Exceeded (1)
16
Data Center Firewall
Rule Lists (Navigate to Security > Network Firewall > Rule Lists)
Rules Continued
Source > Address/Region
Any
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
outbound-snat-nets
Destination > Port
Any
Action
Accept Decisively
Name
no-snats-icmp4
Protocol
ICMP Make sure 1 appears in the box on the right.
ICMP Message
Click Repeat and then add the following rule.
Type
Code
Any
Any
Source > Address/Region
Any
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
outbound-snat-nets
Destination > Port
Any
Action
Drop
Name
ok-others-icmp4
Protocol
ICMP Make sure 1 appears in the box on the right.
ICMP Message
Click Repeat and then add the following rule.
Type
Code (then click Add)
Echo Reply (0)
No Code (0)
Destination Unreachable (3)
Net Unreachable (0)
Destination Unreachable (3)
Host Unreachable (1)
Destination Unreachable (3)
Fragmentation needed and Don’t Fragment… (4)
Destination Unreachable (3)
Communication Administratively Prohibited (13)
Echo Request (8)
No Code (0)
Time Exceeded (11)
Time To Live exceeded in transit (0)
Time Exceeded (11)
Fragment Reassembly Time Exceeded (1)
Source > Address/Region
Any
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
Any
Destination > Port
Any
Action
Accept Decisively
Name
no-unk-icmp4
Protocol
ICMP Make sure 1 appears in the box on the right.
ICMP Message
Source > Address/Region
Any
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
Any
Destination > Port
Any
Action
Drop
Click Repeat and then add the following rule.
Type
Code
Any
Any
Click Finished.
This completes the Network Firewall Rule list configuration.
F5 Deployment Guide
17
Data Center Firewall
Creating the Network Firewall Policies
The next task is to create the Network Firewall policies on the BIG-IP AFM. Use the following tables for guidance on configuring
Network Firewall policies.
Network Firewall Policies (Navigate to Security > Network Firewall > Policies)
Accept All Policy
Description of policy
When AFM is in Firewall mode you must enforce a permissive named firewall policy while you stage a candidate selective policy
on any object (otherwise the implicit policy for that object will discard all packets). The following is the permissive policy.
Policy Name
accept-all
Rules
Click the Add button and add the following rule.
Click Finished and then from the Policies table, click the Policy you just created.
Name
ok-all
Protocol
Any
Source > Address/Region
Any
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
Any
Destination > Port
Any
Action
Accept Click Finished.
Global Policy
Description of policy
The global policy disposes of many erroneous, spoofed, fuzzed, and garbage packets. It also deals with ICMP (In TMOS 11.6
only global and route-domain firewall policies may filter ICMP traffic). Lacking a “VLAN context” you will filter all packets from the
Internet using rules keyed to VLAN external in the global context. This saves adding multiple rules to every public-facing virtual
server’s policy.
Policy Name
global
Rules
Click the Add button and add the following rules
Click Finished and then from the Policies table, click the Policy you just created.
Name
g-icmp
Type
Rule List
Rule List
From the list, under /Common, select the global-icmp-control Rule List you created
Click Repeat and then add the following rule
Name
g-bad-nets
Type
Rule List
Rule List
From the list, under /Common, select the global-forbid-nets Rule List you created
Click Repeat and then add the following rule
Name
g-ctrl-inet
Type
Rule List
Rule List
From the list, under /Common, select the control-from-Internet Rule List you created
Click Finished
Self IP Policy
Description of policy
BIG-IP systems operate best when Port Lockdown is set to Default on all self IP addresses. However, you should apply a firewall
policy to restrict interactive management traffic to the management port. The global policy also protects self IP’s.
Policy Name
global
Rules
Click the Add button and add the following rules
Click Finished and then from the Policies table, click the Policy you just created.
Name
no-mgmt
Type
Rule List
Rule List
From the list, under /Common, select the block-selfIP-mgmt Rule List you created
Click Repeat and then add the following rule
F5 Deployment Guide
Name
ok-others
Protocol
Any
Source > Address/Region
Any
Source > Port
Any
Source > VLAN/Tunnel
Any
Destination > Address/Region
Any
Destination > Port
Any
Action
Accept Click Finished.
18
Data Center Firewall
Network Firewall Policies (Navigate to Security > Network Firewall > Policies)
DHCP relay Policy
Description of policy
Like other listeners, DHCP relay virtual servers must be secured
Policy Name
ok-dhcp
Rules
Click the Add button and add the following rule.
Click Finished and then from the Policies table, click the Policy you just created.
Name
no-mgmt
Type
Rule List
Rule List
From the list, under /Common, select the permit-dhcp Rule List you created
Click Finished
Outbound filter Policy
Description of policy
A single policy will suffice for several “outbound SNAT” virtual servers. The Rule list used in this policy (filter-outbound) includes
a rule to prevent intranet devices sending packets to public multicast addresses. You might have to insert exceptions for some
applications (e.g., a streaming-media server). See the no-send-mcast entry in Creating the Network Firewall Rule list on page
14.
Click Finished and then from the Policies table, click the Policy you just created.
Policy Name
outbound
Rules
Click the Add button and add the following rule.
Name
ob-filter
Type
Rule List
Rule List
From the list, under /Common, select the filter-outbound Rule List you created
Click Finished
This completes the Network Firewall Policy configuration.
F5 Deployment Guide
19
Data Center Firewall
IP Intelligence
The model network architecture uses IP geolocation with network firewall rules to exclude connections from countries on the ITAR list.
Other uses for IP geolocation include blocking regional script kiddies and enforcing copyright license boundaries. As a backstop to
IP geolocation, in this example we configure a global IP Intelligence policy to exclude connections from proxy services that might be
hiding unwelcome source addresses. An F5 IP Intelligence subscription is strongly recommended but not required.
As a key DoS defense we exclude other bad-reputation source addresses, including networks blacklisted by Spamhaus.org, as well
as bogons with help from Team Cymru. To reduce the load on your Internet link and BIG-IP AFM, try persuading your ISP to drop
bogons upstream.
The best practice is to whitelist the public addresses of business partners plus your own public addresses, especially those of site-tosite VPN endpoints. Initially the whitelist will contain 192.0.2.0/24 and 2001:db8:16d::/48 in our example.
Creating the iRule to retrieve address lists
First, we create this iRule to help fetch certain address lists.
From the Main tab, expand Local Traffic and then click iRules. Click the Create button. Give the iRule unique name (we use
afm-feedlist-helper), and then use the following code in the Definition section, omitting the line numbers:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
Team Cymru's bogon list and the Spamhaus.com E+DROP
lists include comments and use addr/masklen rather
than the CSV addr,masklen format the v11.6 Feed List
parser wants. To read those or similar lists (any
with lines that begin "addr/masklen") into AFM:
Create a virtual server with HTTP Profile on a nonrouteable IP like 198.18.0.1:80. Attach this iRule.
No pool. Use a Feed URL source URL like:
http://198.18.0.1/www.spamhaus.org/drop/drop.txt
with the real hostname in the first part of the URL
path. This iRule will proxy requests to the real
host and reformat the response to suit AFM.
Ensure BIG-IP DNS resolver is setup per SOL12224.
when HTTP_REQUEST {
if {([HTTP::method] ne "GET") ||
![regexp {^/([^/]+)(/.*)$} [HTTP::uri] x h u] ||
([set ip [lindex [RESOLV::lookup $h] 0]] eq "")} {
log local0.err "unable to resolve ${h}"
HTTP::respond 502 ; return
}
HTTP::uri $u
HTTP::version "1.0" ; #avert response-chunking
HTTP::header replace Host $h
HTTP::header replace Accept-Encoding identity
HTTP::header replace Connection close
node $ip 80
}
when HTTP_RESPONSE {
if {[HTTP::status] != 200} { return }
if {[set clen [HTTP::header Content-Length]] eq ""} {
set clen 2097152
}
HTTP::collect $clen
}
when HTTP_RESPONSE_DATA {
#discard all but addresses, make those CSV
set r {(?w)(?:(?:(^[0-9A-Fa-f:.]+/[0-9]{1,3})[^\n]*(\n))}
append r {|(?:(?!^[0-9A-Fa-f:.]+/)[^\n]*\n))}
regsub -all $r [HTTP::payload] {\1\2} buf
set buf [string map {/ ,} $buf]
HTTP::payload replace 0 [HTTP::payload length] $buf
HTTP::release
}
F5 Deployment Guide
20
Data Center Firewall
Creating a virtual server for the iRule
The next task is to create a helper virtual server to run the iRule. Create this virtual server separately on each BIG-IP device (big-s1
and big-s2 in our example) and move its virtual address to traffic-group none.
Virtual Servers (Navigate to Local Traffic > Virtual Servers)
Name
Type a unique name, such as feedlist-helper-vs
Type
Standard
Destination Address
198.18.0.1
HTTP Profile
From the list, under /Common, select http.
VLAN and Tunnel Traffic
Select Enabled On, and then select only the Internal VLAN you created.
iRules
Enable the afm-feedlist-helper iRule you created. Do not select a Default pool for this virtual server.
Modifying the Traffic Group
Once you have created the virtual server, you must modify the Traffic Group setting. If necessary, click Local Traffic > Virtual
Servers. On the menu bar, click Virtual Address List. Click the IP address of the virtual server you just created (198.18.0.1 in our
example). From the Traffic Group list, select None, and then click Update.
Creating blacklist categories and feed lists
The next task is to create blacklist categories and then Feed Lists to populate them. Use the following table for guidance.
Black List Categories (Navigate to Security > Network Firewall > IP Intelligence > Black List Categories)
Bogons category
Black List Category Name
bogons
Description
Unallocated or invalid IPv6 prefixes and IPv4 blocks. Click Repeat and then create the following category.
DROP_EDROP category
Black List Category Name
DROP_EDROP
Description
Malevolent source networks. Click Finished
Next, create the associated feed lists using the following table as guidance.
Feed Lists (Navigate to Security > Network Firewall > IP Intelligence > Feed Lists)
Bogons Feed List
Name
bogons-feed
Description
bogons courtesy of Team Cymru
Feed List Properties: Feed URLs
Name
bogons-6
URL
http://198.18.0.1/www.team cymru.org/Services/ Bogons/fullbogons-ipv6.txt
List Type
Black List
Blacklist Category
bogons
Poll Interval
14400 Click Add and then create the following Feed URL.
Name
bogons-4
URL
http://198.18.0.1/www.team cymru.org/Services/ Bogons/fullbogons-ipv4.txt
List Type
Black List
Blacklist Category
bogons
Poll Interval
14400 Click Add and then click Finished.
DROP_EDROP feed Feed List
Name
DROP_EDROP
Description
Don’t Route Or Peer courtesy of Spamhaus.org
Feed List Properties: Feed URLs
Name
DROP
URL
http://198.18.0.1/www.spamhaus.org/drop/drop.txt
List Type
Black List
Blacklist Category
DROP_EDROP
Poll Interval
86400 Click Add and then create the following Feed URL.
F5 Deployment Guide
21
Data Center Firewall
Feed Lists (Navigate to Security > Network Firewall > IP Intelligence > Feed Lists)
Feed List Properties: Feed URLs
Name
EDROP
URL
http://198.18.0.1/www.spamhaus.org/drop/edrop.txt
List Type
Black List
Blacklist Category
DROP_EDROP
Poll Interval
86400 Click Add and then click Finished.
Creating the Local whitelist feed
Another Feed List is required to populate the IP Intelligence policy’s whitelist. Before you create it, identify a web or FTP server to
provide the whitelist addresses and note the corresponding URL so you can point a Feed URL at it. If you lack a suitable server, see
Appendix B: Storing IP Intelligence Address Lists in BIG-IP Data Groups on page 62 for a workaround .
Feed Lists (Navigate to Security > Network Firewall > IP Intelligence > Feed Lists)
Name
bogons-feed
Description
Trusted addresses for our company and our business partners.
Feed List Properties: Feed URLs
Name
local-whitelist-feed
URL
URL of the Web or FTP server. If you do not have one, see Appendix X on page 62
List Type
White List
Blacklist Category
spam_sources (What you choose here does not matter; it can be anything)
Poll Interval
600
Creating an IP Intelligence policy
The final task for IP Intelligence is to create a policy that uses the objects you created in this section. Even if your BIG-IP system does
not have an F5 IP Intelligence subscription, you can still use custom Feed Lists and black list categories.
IP Intelligence Policy (Navigate to Security > Network Firewall > IP Intelligence > Policies)
Name
global-IPI-policy
Description
Prevents connections from bogus and malicious source addresses
IP Intelligence Properties
Feed Lists
Use the Add arrows (<<) to move the three Feed Lists you created (bogons-feed,
DROP_EDROP-feed, and local-whitelist-feed) to the Selected box.
Default Action
Drop
Default Log Actions
Check the Log Blacklist Category Matches and Log Whitelist Overrides boxes.
Blacklist Matching Policy
From the Blacklist Category list, select each of the following categories and then
click Add (leave the Action and both log lists at the default (Use Policy Default)
bogons
botnets
denial_of_service
DROP_EDROP
illegal_websites
infected_sources
phishing
proxy
scanners
spam_sources
web_attacks
windows_exploits
F5 Deployment Guide
22
Data Center Firewall
Configuring AFM High-Speed Logging
The AFM sends log messages at high speed to a log server. Use this section to configure the objects for high-speed logging.
Pools (Navigate to Local Traffic > Pools)
Name
afm-log-pool
Health Monitor
Select gateway_icmp.
New Members
Node Name
logfarm
Node Address
fdf5:f:5:c006::17
Service Port
514
Log Destination (Navigate to System > Logs > Configuration > Log Destinations)
Remote High-Speed log destination
Name
afm-log-dest
Type
Remote High-Speed Log
Pool Name
Select the pool you created (afm-log-pool in our example)
Protocol
UDP
Remote Syslog destination
Name
afm-syslog-dest
Type
Remote Syslog
Syslog Format
BSD Syslog
High-Speed Log Destination
afm-log-dest
Log Publisher (Navigate to System > Logs > Configuration > Log Publishers)
Name
afm-log-pub
Destinations
From the Available box, select the afm-syslog-dest destination and then use the Add arrows (<<) to move it to the Selected box.
Logging Profile (Navigate to Security > Event Logs > Logging Profiles)
Profile Name
afm-log-pfl
Protocol Security
Check the box to Enable Protocol Security.
Network Firewall
Check the box to Enable Network Firewall.
DoS Protection
Check the box to Enable DoS Protection.
Protocol Security Tab
Click the Protocol Security tab.
HTTP, FTP, and SMTP Security: Publisher
Network Firewall Tab
DoS Protection Tab
Click the Network Firewall tab.
Network Firewall Publisher
Select the Log Publisher you created (afm-log-pub in our example).
Log Rule Matches
Click the Drop and Reject boxes. Leave the Rate Limit set to Indefinite
Log Translation Fields
Check the box to Enable Log Translation Fields
Storage Format
Select Field List, and leave Delimiter set to a comma (,)
From the Available Items list, select the following formats and use the Add (<<)
button to move them to the Selected Items list in the following order:
date_time
bigip_hostname
context_name
acl_policy_name
acl_rule_name
action
protocol
vlan
src_ip
src_port
dest_ip
dest_port
drop_reason
source_user
translated_vlan
translated_src_port
translated_src_ip
translated_dest_port
translated_dest_ip
IP Intelligence Publisher
Select the Log Publisher you created (afm-log-pub in our example).
Click the DoS Protection tab.
Network DoS Protection: Publisher
F5 Deployment Guide
Select the Log Publisher you created (afm-log-pub in our example).
Select the Log Publisher you created (afm-log-pub in our example).
23
Data Center Firewall
Global DoS Protection
The Dos Protection Device Configuration (global) policy is always enforced. As explained in Understanding AFM DoS Protection on
page 59, you must analyze your application traffic before adjusting this policy. Here you simply enable logging to permit data to
accumulate for analysis.
DoS Protection Log Publisher (Navigate to Security > DoS Protection > Device Configuration)
Log Publisher
From the list, select the Log Publisher you created (afm-log-publisher in our example). Click Update.
Preparing for outbound traffic
For outbound access through the data center firewall, use the following tables for guidance on configuring the following objects.
Creating the SNAT Pool
A SNAT pool represents a pool of translation addresses that you configure on the BIG-IP system. Use the following table for guidance
on configuring a SNAT pool.
SNAT Pool (Navigate to Local Traffic > Address Translation > SNAT Pool List)
Name
Type a unique name. In our example, we use outbound-snatpool
Member List: IP Address
Type the appropriate SNAT addresses, clicking the Add button after each. In our example, we type:
192.0.2.240
192.0.2.241
192.0.2.242
192.0.2.243
2001:db8:16d:2::240
2001:db8:16d:2::241
2001:db8:16d:2::242
2001:db8:16d:2::243
Configuring the BIG-IP system for outbound FTP
Use the following table for guidance on configuring the BIG-IP system for outbound FTP.
Pools (Navigate to Local Traffic > Pools)
IPv4 Gateway pool
Name
Type a unique name. In our example, we use internet-gw-4-pool
Health Monitor
Select gateway_icmp.
New Members
Node Name
internet-gw-4
Node Address
192.0.2.1
Service Port
*All Services (0)
IPv6 Gateway pool
Name
Type a unique name. In our example, we use internet-gw-6-pool
Health Monitor
Select gateway_icmp.
New Members
Node Name
internet-gw-6
Node Address
2001:db8:16d:2::1
Service Port
*All Services (0)
Virtual Servers (Navigate to Local Traffic > Virtual Servers)
IPv4 FTP Gateway virtual server
Name
ftp-outbound-4-vs
Type
Standard
Destination Address
0.0.0.0/0
Service Port
21 (FTP)
FTP Profile
From the list, under /Common select ftp.
VLAN and Tunnel Traffic
Select Enabled On, and then select only the Internal VLAN you created.
Connection Mirroring
Check the box to Enable Connection Mirroring. You must select Advanced from the list at the top of the Configuration area.
Source Address Translation
SNAT
SNAT Pool
Select the SNAT pool you created (outbound-snatpool in our example).
Default Pool
Select the IPv4 gateway pool you created (internet-gw-4-pool in our example).
F5 Deployment Guide
24
Data Center Firewall
IPv6 FTP Gateway virtual server
Name
ftp-outbound-6-vs
Type
Standard
Destination Address
::/0
Service Port
21 (FTP)
FTP Profile
From the list, under /Common select ftp.
VLAN and Tunnel Traffic
Select Enabled On, and then select only the Internal VLAN you created.
Connection Mirroring
Check the box to Enable Connection Mirroring. Select Advanced from Configuration list (top of pane) for this option to appear.
Source Address Translation
SNAT
SNAT Pool
Select the SNAT pool you created (outbound-snatpool in our example).
Default Pool
Select the IPv4 gateway pool you created (internet-gw-6-pool in our example).
Configuring the BIG-IP system for outbound TCP
Use the following table for guidance on configuring the BIG-IP system for outbound TCP traffic.
Rather than mirror outbound TCP connections, we enable Loose Initiation and Loose Close, and then disable Reset on Timeout in a
Fast L4 Profile (see SOL7595). This averts loss of end-user connectivity in the event of a BIG-IP Traffic Group failover. Note that you
typically use connection mirroring with application layer gateways (ALGs) like the outbound FTP virtual servers.
Fast L4 Profiles (Navigate to Local Traffic > Profiles > Protocol > FastL4)
TCP Loose Fast L4 Profile
Name
Type a unique name. In our example, we use tcp-loose-fastl4
Reset on Timeout
Clear the box to Disable Reset on Timeout
TCP Handshake Timeout
Leave Specify selected, and then in the Seconds box, type 10
Loose Initiation
Check the box to Enable Loose Initiation
Loose Close
Check the box to Enable Loose Close
Virtual Servers (Navigate to Local Traffic > Virtual Servers)
IPv4 FTP Gateway virtual server
Name
tcp-outbound-4-vs
Type
Forwarding (IP)
Destination Address
0.0.0.0/0
Service Port
* (* All ports)
Protocol
TCP
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the Internal VLAN you created.
Source Address Translation
SNAT
SNAT Pool
Select the SNAT pool you created (outbound-snatpool in our example).
IPv6 FTP Gateway virtual server
Name
tcp-outbound-6-vs
Type
Forwarding (IP)
Destination Address
::/0
Service Port
* (* All ports)
Protocol
TCP
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the Internal VLAN you created.
Source Address Translation
SNAT
SNAT Pool
Select the SNAT pool you created (outbound-snatpool in our example).
F5 Deployment Guide
25
Data Center Firewall
Configuring the BIG-IP system for outbound ICMP and UDP
To pass ICMP through the BIG-IP (so IPv6 will work and so intranet users can ping Internet resources) you need to create virtual
servers that support the protocol *All Protocols. The simplest approach is to let those handle UDP as well, and block other
protocols using the AFM network firewall policy.
Use the following table for guidance on configuring the BIG-IP system for outbound ICMP traffic.
Fast L4 Profiles (Navigate to Local Traffic > Profiles > Protocol > FastL4)
Name
Type a unique name. In our example, we use udp-outbound-fastl4
Idle Timeout
Leave Specify selected, and then in the Seconds box, type 15
Virtual Servers (Navigate to Local Traffic > Virtual Servers)
IPv4 ICMP Outbound virtual server
Name
udp-icmp-outbound-4-vs
Type
Forwarding (IP)
Destination Address
0.0.0.0/0
Service Port
* (* All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the Fast L4 profile you created (udp-outbound-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the Internal VLAN you created.
Source Address Translation
SNAT
SNAT Pool
Select the SNAT pool you created (outbound-snatpool in our example).
IPv6 FTP Gateway virtual server
Name
udp-icmp-outbound-6-vs
Type
Forwarding (IP)
Destination Address
::/0
Service Port
* (* All ports)
Protocol
TCP
Protocol Profile (Client)
Select the Fast L4 profile you created (udp-outbound-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the Internal VLAN you created.
Source Address Translation
SNAT
SNAT Pool
Select the SNAT pool you created (outbound-snatpool in our example).
Activating BIG-IP AFM Firewall Mode
When AFM is in Firewall mode, no virtual server or self IP processes traffic unless a sufficiently-permissive network firewall policy is
enforced on it. Use the following table for guidance on activating Firewall mode. When you activate Firewall mode, remind all BIG-IP
administrators to consider the presence or absence of firewall policies when they troubleshoot problems with BIG-IP virtual servers.
Default Firewall Action (Navigate to Security > Options > Network Firewall)
Virtual Server & Self IP Contexts
Select Drop
Global Context
Select Drop, and then click Update.
F5 Deployment Guide
26
Data Center Firewall
Applying Fundamental Firewall Protection
The next task is to apply fundamental firewall protection on the BIG-IP AFM.
First, navigate to Security > Network Firewall > Active Rules. Click the word Global at the start of the first line of the list. A new
page opens.
Figure 2:
Global link on the Active Rules page
Use the following table for guidance on configuring the global firewall rules.
Global Firewall Rules (Navigate to Security > Network Firewall > Active Rules > 'Global' link)
Network Firewall: Enforcement
Click Enabled From the Policy list that appears, select the global policy you created in Creating the Network Firewall
Policies on page 18 (global in our example).
Network Firewall: Staging
Leave Staging set to Disabled. Click Update.
The next task is to attach a firewall policy to the self IP addresses in the local-only traffic group. This needs to be performed on each
BIG-IP device (big-s1 and big-s2) in our example. Then, on the primary device, attach the same policy to the self IPs in traffic-group-1
(floating), and use config sync to update the other device.
You may need to briefly enforce a different policy such as /Common/accept-all on one self IP while you configure a management-port
firewall policy. See Protecting the BIG-IP Management Port with AFM on page 46.
Self IPs (Navigate to Network > Self IPs)
Navigation help
Click Network > Self IPs. Click one of the self IPs in the traffic-group-local-only (use the Traffic Group column for
reference). Once on the self IP properties page, click the Security tab.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the self IP policy you created in Creating the Network Firewall
Policies on page 18 (self IP in our example).
Network Firewall: Staging
Leave Staging set to Disabled. Click Update.
Repeat this guidance for each of the self IPs in traffic-group-local-only. Then repeat all on the second BIG-IP system.
Finally, repeat this guidance on the primary device to the self IPs in traffic-group1, and then use configsync to update
the other device. See https://support.f5.com/kb/en-us/products/big-ip_ltm/manuals/product/bigip-device-serviceclustering-admin-11-6-0/5.html#unique_9477342 for specific information on syncing the BIG-IP configuration.
Next, on the Security tab of your DHCP relay virtual server(s), attach the firewall policy as shown in the following table. After testing
and log analysis you can move the policy from staging to enforcement.
F5 Deployment Guide
27
Data Center Firewall
DHCP virtual server (Navigate to Local Traffic > Virtual Servers > your-DHCP-relay-virtual-server > Security > Policies)
Navigation help
Navigate to Local Traffic > Virtual Servers. From the Virtual Server list, click the DHCP relay virtual server you
created in Configuring a DHCPv6 Relay on page 10 (or similar for IPv4 DHCP). On the Virtual Server Properties page,
from the Menu bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network
Firewall Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the DHCP relay policy you created in Creating the Network
Firewall Policies on page 18 (ok-dhcp) in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
If you also created an IPv4 DHCP relay virtual server, repeat this guidance for the IPv4 virtual server.
Applying a global IP Intelligence policy
The next task is to apply a global IP Intelligence policy to the configuration.
Global IP Intelligence Policy (Navigate to Security > Network Firewall > IP Intelligence > Policies)
IP Intelligence Policy
From the list, select the IP Intelligence policy you created in Creating an IP Intelligence policy on page 22
(global-IPI-policy in our example)
Click Update.
Securing Outbound Traffic
This section provides guidance on securing outbound traffic through the BIG-IP system. In this example, we show how to stage the
initial policies. After you generate some test traffic and make any adjustments suggested by log analysis you can move your policies
from staging to enforcement.
Virtual Server (Navigate to Local Traffic > Virtual Servers > your-outbound-virtual-servers > Security > Policies )
Navigation help
Click Local Traffic > Virtual Server. Click one of the virtual servers you created for outbound traffic. On the Menu
bar, click Security > Policies. In our example, the outbound virtual servers are: ftp-outbound-4-vs, ftp-outbound-6-vs,
tcp-outbound-4-vs, tcp-outbound-6-vs, udp-icmp-outbound-4-vs, and udp-icmp-outbound-6-vs.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network
Firewall Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the outbound filter policy you created in Creating the Network
Firewall Policies on page 18 (outbound in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
Repeat this guidance for each of the outbound virtual servers.
Don’t forget to set the BIG-IP floating self IP addresses fdf5:f:5:c245::1 and 10.1.245.1 (VLAN “internal”) as the default gateway
addresses on the data center routers (or L3 switches).
F5 Deployment Guide
28
Data Center Firewall
Additional information: Remote AFM Logging
You must provide a remote log server to accept log messages from AFM. The BIG-IP system supports several log formats. Our
example uses the most common—syslog.
Most organizations capture detailed logs to support alerting, event correlation, incident management, security forensics, policy
debugging, and so forth. AFM can emit many log messages very quickly. AFM uses the BIG-IP High Speed Logging (HSL) feature. In
TMOS v11 High Speed Logging messages from AFM can only leave the BIG-IP system on traffic-handling VLAN's (see SOL14459).
However, TMOS v12 added the ability to send High Speed Logging messages out the management port, though at a serious
performance cost. You must create a Log Destination of type "Management Port" then enter the IP address, port, and protocol of the
destination log server. After you create the Log Destination you can reference it in a Log Publisher.
(In default of a suitable logserver, you can simply disable logging most of the time but use the BIG-IP device’s internal log storage to
capture short bursts of log messages while testing new policies or troubleshooting network problems. That is not a recommended
posture.)
Consider connecting your BIG-IP devices to a special logging VLAN/subnet to avoid mixing application and logging traffic. Mixing
traffic can lead to congestion and delays. Another reason to segregate logging traffic is to bypass router and EW firewall hops in your
intranet: if you send log messages directly onto the subnet where your log/SIEM server(s) live you may gain a lot of performance.
Additional Information: Internet Access from the Intranet
All traffic from the intranet to the Internet (NS) requires address translation as well as security filtering. You SNAT the private (intranet)
source addresses of outbound connections using a pool of public addresses. To make your firewall rules more efficient use SNAT
addresses you can summarize in CIDR blocks, as shown in our example. SNAT also helps hide details of the intranet from malicious
users.
To protect intranet devices and users against various threats and to frustrate APT’s trying to “phone home” you should prevent
outbound connections to addresses blacklisted by the IP Intelligence service. With TMOS v11.6 this requires an iRule like afm-noevil-dst which we define in the example. To keep APT’s from sending spam it is customary to forbid connections to TCP port 25
(SMTP) except from the organization’s mailserver.
More elaborate security policies may be desirable. You can add rules to keep intranet devices from accessing Internet services
improperly. For example, you might forbid outbound connections to TCP port 179 (BGP) to prevent APT’s harassing remote
organizations.
Most application protocols use TCP and UDP in a straightforward manner but some like FTP require extra support (an ALG) to pass
through the BIG-IP. You set up “more specific” virtual servers to process such protocols. For example, the basic configuration
includes specific virtual servers to support outbound FTP connections as explained in SOL8021: Configuring the BIG-IP LTM system
to allow outbound FTP sessions. Deploying other BIG-IP ALG’s is basically similar.
An organizational “web proxy” is sort of “the ultimate ALG.” A typical web proxy—such as BIG-IP Secure Web Gateway (SWG)—
does a lot of security work. It enforces the organization’s Acceptable Use Policy (AUP), guards against malicious web content, scans
downloads for malware, and much more besides. Though specific details are beyond the scope of this document, to integrate your
web proxy with your BIG-IP data center firewall you will create wildcard virtual servers on TCP ports 80 and 443 (at least) to collect
traffic “transparently” from users and send it to the proxy. You will also create some source-specific virtual servers to pass the webproxy’s outbound traffic to the Internet.
You may also integrate Data Loss Prevention (DLP) or similar services with your BIG-IP data center firewall. The BIG-IP system
supports ICAP for simple use cases, and much more comprehensive IDS/IPS/NGFW integrations when desired.
F5 Deployment Guide
29
Data Center Firewall
Securing an Email Gateway
A typical organization has a spam-filtering email gateway to handle SMTP inbound and outbound. The model network includes a mail
server trusted to operate in the intranet rather than confined to a DMZ. It is very secure and uses DNS real-time blacklists to reject
messages which come from undesirable senders or include dangerous links. Accordingly, the mail server needs to see the actual
source IP address of every incoming SMTP connection. It is dual-stacked to support both IPv6 and IPv4 addresses conveniently.
Of course F5 customers use a variety of email solutions. Microsoft and F5 are technology partners and support a powerful integration
of Microsoft Exchange Server with the BIG-IP system. See the F5 Microsoft Exchange Deployment Guide for information and details
on the iApp template for Microsoft Exchange Client Access Servers.
Though the example in this guide is simple, the F5 document Deploying the BIG-IP System with SMTP Servers offers guidance for a
variety of usage scenarios. For instance, it explains how to offload TLS processing for SMTP from mail servers to the BIG-IP system
by attaching an SMTPS Profile and a Client SSL Profile to each SMTP virtual server.
Create the following objects to support inbound SMTP traffic (the basic AFM configuration handles outbound SMTP in the Rule List
filter-outbound). This configuration requires two pools in order to pass both IPv6 and IPv4 remote addresses through to the mail
server.
Network Firewall Policies (Navigate to Security > Network Firewall > Policies)
Description of policy
This network firewall policy is for inbound email. The Destination address “any” covers IPv6 and IPv4, and lets you adjust
virtual server addresses when, for example, you change ISP’s and get a new public IP block.
Policy Name
inbound-mail
Rules
Click the Add button and add the following rule.
Click Finished and then from the Policies table, click the Policy you just created.
Name
ok-smtp
Protocol
TCP
Source > Address/Region
Any
Source > Port
Any
Source > VLAN/Tunnel
external
Destination > Address/Region
Any
Destination > Port
smtp-mta
Action
Accept Click Finished.
Monitors (Navigate to Local Traffic > Monitors)
Name
mon-smtp
Type
SMTP
Interval
30
Timeout
91
Domain
Type the appropriate domain. In our example, we use example.net.
Pools (Navigate to Local Traffic > Pools)
IPv4 mail server pool
Name
Type a unique name. In our example, we use mailserver-4-pool
Health Monitor
mon-smtp
New Members
Node Name
mailserver-4
Node Address
10.1.7.25
Service Port
25 (SMTP)
IPv6 mail server pool
Name
Type a unique name. In our example, we use mailserver-6-pool
Health Monitor
mon-smtp
New Members
F5 Deployment Guide
Node Name
mailserver-6
Node Address
fdf5:f:5:c007::25
Service Port
25 (SMTP)
30
Data Center Firewall
Virtual Servers (Navigate to Local Traffic > Virtual Servers)
IPv4 mail server virtual server
Name
inbox-4-vs
Type
Standard
Destination Address
192.0.2.25
Service Port
25 (SMTP)
VLAN and Tunnel Traffic
Select Enabled On, and then select only the external VLAN you created.
Default Pool
Select the IPv4 mail server pool you created (mailserver-4-pool in our example).
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network Firewall
Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the inbound mail policy you just created (inbound-mail in our
example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on page
23 (afm-log-pfl in our example).
IPv6 mail server virtual server
Name
inbox-6-vs
Type
Standard
Destination Address
2001:db8:16d:2::25
Service Port
25 (SMTP)
VLAN and Tunnel Traffic
Select Enabled On, and then select only the external VLAN you created.
Default Pool
Select the IPv4 mail server pool you created (mailserver-6-pool in our example).
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network Firewall
Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the inbound mail policy you just created (inbound-mail in our
example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on page
23 (afm-log-pfl in our example).
F5 Deployment Guide
31
Data Center Firewall
Securing a Multi-Tier Web Application with Presentation Servers in a DMZ
DMZ’s hold servers which mediate between high- and low-trust environments. A DMZ typically has two trust boundaries (NS and
EW), policed by firewall policy but also enforced by the LTM configuration. Among other things, the EW trust boundary should keep
an adversary who compromises a server in the DMZ from extending his intrusion into the intranet.
This example shows a multi-tier web application (App1) with presentation servers in the DMZ (DMZ1) and additional (e.g., businesslogic/database) processing in the data center. Only two kinds of traffic may enter DMZ1: web traffic and system-administration
traffic. Most of the web traffic comes from the Internet but some comes from the intranet. Sys-admin traffic comes only from the
management network in the intranet. Two kinds of traffic exit DMZ1: logging and limited admin (e.g., DHCP) traffic plus application
traffic only to the application backend service (this might be a BIG-IP virtual server with a pool of application servers behind it). DMZ1
machines may initiate connections to approved data-center resources but nothing else.
In this section, the first task is to configure IP forwarding virtual servers with a network firewall policy. They carry and control DMZ1
EW traffic for administration, logging, and backend access. The next step is to deploy application-specific virtual servers, which admit
NS and EW web traffic to DMZ1. The web virtual servers let you use an AFM HTTP Protocol Security Profile to protect the application
and support server load balancing.
In this example App1 presentation servers talk SOAP or REST over HTTPS to their backend. To support another protocol, you could
change the port in the relevant firewall rule (Rule “ok webback1”, part of rule-list “permit dmz1-to dc") from https (443) to something
like ms-sql (1433) for MS SQL Server 2012.
Use the following table for guidance on configuring the objects on the BIG-IP system. For instructions on configuring individual
objects, see the Help tab or the product documentation.
VLANs (Navigate to Network > VLANs)
Notes
Create this VLAN on each device. It carries the first DMZ subnet.
Name
Type a unique name. In our example, we use dmz1. Configure tagging as appropriate for your deployment.
Self IPs (Navigate to Network > Self IPs)
On the big-s1 device, create the following Self IPs in the /Common partition
Name
Type a unique name. In our example, we use self-dmz1-s1-4
IP Address
172.30.1.2
Netmask
255.255.255.0
VLAN
Select the VLAN you created in this section (dmz1 in our example)
Port Lockdown
Allow Default
Traffic Group
traffic-group-local-only (non-floating)
Name
Type a unique name. In our example, we use self-dmz1-s1-6
IP Address
fdf5:f:5:d001::2
Netmask
ffff:ffff:ffff:ffff::
VLAN
Select the VLAN you created in this section (dmz1 in our example)
Port Lockdown
Allow Default
Traffic Group
traffic-group-local-only (non-floating)
On the big-s2 device, create the following Self IPs in the /Common partition
Name
Type a unique name. In our example, we use self-dmz1-s2-4
IP Address
172.30.1.3
Netmask
255.255.255.0
VLAN
Select the VLAN you created in this section (dmz1 in our example)
Port Lockdown
Allow Default
Traffic Group
traffic-group-local-only (non-floating)
Name
Type a unique name. In our example, we use self-dmz1-s2-6
IP Address
fdf5:f:5:d001::3
Netmask
ffff:ffff:ffff:ffff::
VLAN
Select the VLAN you created in this section (dmz1 in our example)
Port Lockdown
Allow Default
Traffic Group
traffic-group-local-only (non-floating)
F5 Deployment Guide
32
Data Center Firewall
On the both devices, create the following floating Self IPs in the /Common partition
Name
Type a unique name. In our example, we use self-dmz1-float-4
IP Address
172.30.1.1
Netmask
255.255.255.0
VLAN
Select the VLAN you created in this section (dmz1 in our example)
Port Lockdown
Allow Default
Traffic Group
traffic-group-1 (floating)
Name
Type a unique name. In our example, we use self-dmz1-float-6
IP Address
fdf5:f:5:d001::1
Netmask
ffff:ffff:ffff:ffff::
VLAN
Select the VLAN you created in this section (dmz1 in our example)
Port Lockdown
Allow Default
Traffic Group
traffic-group-1 (floating)
Configuring DMZ1 Administrative and Backend Traffic
Note: You may download an iApp to perform much of the following configuration from:
https://devcentral.f5.com/codeshare/data-center-firewall-quick-start-iapp-template
On each BIG-IP device, issue one of the following TMSH commands, depending on your configuration.
In the model network architecture, servers learn DNS settings from DHCPv6, so the command is:
create /net router-advertisement ra-dmz1 vlan dmz1 enabled prefixes
add { /Common/pfx-dmz1 { prefix fdf5:f:5:d001:: prefix-length 64 }
router } managed other-config
If you set up a network without DHCPv6, substitute this command:
create /net router-advertisement ra-dmz1 vlan dmz1 enabled prefixes
add { /Common/pfx-dmz1 { prefix fdf5:f:5:d001:: prefix-length 64 }
router }
Save your changes using the following command:
save /sys config
Configuring the BIG-IP objects
Use the following table for guidance on configuring the BIG-IP system.
DHCP virtual server (Navigate to Local Traffic > Virtual Servers > your-DHCP-relay-virtual-server)
Navigation help
Navigate to Local Traffic > Virtual Servers. From the Virtual Server list, click the DHCP relay virtual server you
created in Configuring a DHCPv6 Relay on page 10 (or similar for IPv4 DHCP).
VLAN and Tunnel Traffic
Add the DMZ VLAN you created (dmz1 in our example) to the Selected list. There should be only two VLANs in the
Selected list, the DMZ VLAN and the Internal VLAN.
If applicable, repeat for the other DHCP relay virtual server.
Address Lists (Navigate to Security > Network Firewall > Address Lists)
Name
all-dmz-nets
Description
(example)
Notes (not a part of the configuration)
Addresses/Regions
Type or copy and paste the following addresses into
the Add new entry box:
172.30.0.0/16
F5 Deployment Guide
All DMZ subnets (part of intranet-nets).
fdf5:f:5:d000::/54
33
Data Center Firewall
Name
dmz1-nets
Description
(example)
Notes (not a part of the configuration)
Addresses/Regions
Type or copy and paste the following addresses into
the Add new entry box:
172.30.1.0/24
DMZ1
fdf5:f:5:d001::/64
Rule Lists (Navigate to Security > Network Firewall > Rule Lists)
permit-mgmt-to-dmz Rule list
Click Finished and then from the Rule Lists table, click the Rule List you just created.
Rule List Name
permit-mgmt-to-dmz
Rules
Click the Add button and add the following rules.
Name
ok-tcp-mgmt
Protocol
TCP
Source > Address/Region
Add net-mgmt-nets
Source > Port
any
Source > VLAN/Tunnel
any
Destination > Address/Region
Add all-dmz-nets
Destination > Port
tcp-EW-generic
Action
Accept
Name
ok-udp-mgmt
Protocol
UDP
Source > Address/Region
Add net-mgmt-nets
Source > Port
any
Source > VLAN/Tunnel
any
Destination > Address/Region
Add all-dmz-nets
Destination > Port
Add snmp-mgr and udp-dhcp
Action
Accept
Click Repeat and then add the following rule.
Click Finished
Use the following table for guidance on configuring the following Rule lists.
Rule List Name
permit-dmz1-to-dc
access-to-webapps
1
Rule Name
Protocol
Source address
Src. Port
Src. VLAN
Dst. Address
Dst. Port
Action
ok-basic-udp
UDP
dmz1-nets
Any
dmz1
all-dc-nets
udp-EW-basic
Accept
ok-basic-tcp
TCP
dmz1-nets
Any
dmz1
all-dc-nets
tcp-EW-basic
Accept
ok-webback1
TCP
dmz1-nets
Any
dmz1
fdf5:f:5:c001:31
HTTPS
Accept
ok-intra
TCP
intranet-nets
Any
internal
Any
Any
Accept
no-ITAR
1
TCP
ITAR-countries
Any
Any
Any
Any
Drop
ok-others
TCP
Any
Any
Any
Any
snmp-mgr
Accept
For the no-ITAR rule list, from the Logging row, select Enabled
Network Firewall Policies (Navigate to Security > Network Firewall > Policies)
Description of policy
This network firewall policy is for the Management network to any DMZ.
Policy Name
mgmt-to-dmz
Rules
Click the Add button and add the following rule.
Click Finished and then from the Policies table, click the Policy you just created.
Name
mgmt2dmz
Type
Rule List
Rule List
permit-mgmt-to-dmz
Description of policy
This network firewall policy is for DMZ1 to the data center.
Policy Name
dmz1-to-dc
Rules
Click the Add button and add the following rule.
F5 Deployment Guide
Click Finished and then from the Policies table, click the Policy you just created.
Name
dmz1-dc
Type
Rule List
Rule List
permit-dmz1-to-dc
34
Data Center Firewall
Virtual Servers (Navigate to Local Traffic > Virtual Servers)
Note the first two objects support management traffic into all DMZ’s. The latter two are DMZ1-specific. Essentially, we trust the management network a lot and
various DMZ’s very little.
IPv4 management to DMZ virtual server
Name
mgmt-to-dmz-4-vs
Type
Forwarding (IP)
Destination Address
172.30.0.0/16
Service Port
* (*All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the internal VLAN you created.
Source Address Translation
None
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
Next, we stage the initial policies. After you generate some test traffic and make any adjustments suggested by log
analysis you can move your policies from staging to enforcement.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network
Firewall Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the management to DMZ policy you just created (mgmt-todmz in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
IPv6 mail server virtual server
Name
mgmt-to-dmz-6-vs
Type
Forwarding (IP)
Destination Address
fdf5:f:5:d000::/54
Service Port
* (*All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the internal VLAN you created.
Source Address Translation
None
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network
Firewall Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the management to DMZ policy you just created (mgmt-todmz in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
IPv6 mail server virtual server
Name
dmz1-to-intraA-4-vs
Type
Forwarding (IP)
Destination Address
10.0.0.0/8
Service Port
* (*All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the DMZ VLAN you created (dmz1 in our example).
Source Address Translation
None
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
On the Menu Bar, click Security > Policies.
F5 Deployment Guide
35
Data Center Firewall
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network
Firewall Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the management to DMZ policy you just created (dmz1-todc in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
IPv6 mail server virtual server
Name
dmz1-to-intraC-4-vs
Type
Forwarding (IP)
Destination Address
192.168.0.0/16
Service Port
* (*All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the DMZ VLAN you created (dmz1 in our example).
Source Address Translation
None
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network
Firewall Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the management to DMZ policy you just created (dmz1-todc in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
IPv6 mail server virtual server
Name
dmz1-to-intra-6-vs
Type
Forwarding (IP)
Destination Address
fdf5:f:5::/48
Service Port
* (*All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the DMZ VLAN you created (dmz1 in our example).
Source Address Translation
None
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network
Firewall Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the management to DMZ policy you just created (dmz1-todc in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
F5 Deployment Guide
36
Data Center Firewall
Creating a Protocol Security Profile
Use the following table for guidance on configuring the Protocol Security profile.
Protocol Security Profiles (Navigate to Security > Protocol Security > Security Profiles > HTTP)
Name
Type a unique name. In our example, we use afm-http-secy-pfl
Parent Profile
http_security
HTTP Protocol Checks Tab
HTTP Protocol Checks
Check the High ASCII characters in headers box. Leave the other settings at the default.
Blocking Page Tab
Response Type
Select Custom Response from the list (you may have to click the Custom box on the right).
Response Headers
HTTP/1.1 200 OK
Content-Type: text/html; charset=ISO-8859-1
Cache-Control: no-cache
Pragma: no-cache
Connection: close
Response Body
Leave the default Response Body.
Configuring the BIG-IP system to send web application traffic into the DMZ
To admit web traffic to the App1 presentation servers in DMZ1, the easiest way to configure the BIG-IP system is to use the
iApp template for HTTP Applications (f5.http) as explained in the F5 Deployment Guide Deploying the BIG-IP System with HTTP
Applications. The first configuration scenario in that Guide—using BIG-IP as a reverse (or inbound) proxy—describes the goal here.
The HTTP iApp template (and the manual configuration described in the deployment guide) supports BIG-IP AFM, including an option
called F5's recommended AFM configuration. That is not actually the AFM configuration recommended in this Deployment Guide,
so you should create an AFM network firewall policy manually using the following guidance. You can then select your policy in the
iApp template.
This example assumes you have an RSA or ECC private key and corresponding public-key TLS certificate signed by a well-known
CA (Certificate Authority) for App1. Before creating any other objects, import the TLS certificate and key as explained in SOL14620:
Managing SSL certificates for BIG-IP systems using the Configuration utility. Name the certificate and key 20YY-app1.example.net
where 20YY is the year the certificate expires. Also import any intermediate-CA certificate(s) (in a bundle if more than one).
The web presentation servers (webfront1, webfront2 in our example) in DMZ1 can use TLS certificates signed by your organization’s
private CA. Import the private CA’s root certificate to the BIG-IP system. Name it something like 2025-Private-CA-Root-example.net.
Use the following table for guidance on configuring the BIG-IP system for web application traffic into the DMZ. After creating these
objects, you will run the iApp template.
Network Firewall Policies (Navigate to Security > Network Firewall > Policies)
Click Finished and then from the Policies table, click the Policy you just created.
Policy Name
app1-access
Rules
Click the Add button and add the following rule.
Name
access-to-app1
Type
Rule List
Rule List
access-to-webapps
Nodes (Navigate to Local Traffic > Nodes)
Name
Type a unique name. We use webfront1.
Address
Type the address. We use fdf5:f:5:d001::41.
Click Repeat and then add the following node
Name
Type a unique name. We use webfront2.
Address
Type the address. We use fdf5:f:5:d001::42.
Click Finished
F5 Deployment Guide
37
Data Center Firewall
iRules (Navigate to Local Traffic > iRules)
Important: H
STS (HTTP Strict Transport Security; rfc6797) is an important security measure for TLS-protected (HTTPS) websites. With BIG-IP TMOS version
11.x an iRule is needed to implement HSTS. BIG-IP TMOS v12.x has support for HSTS built in (though the iRule will still work—use it when you want
to defer to HSTS headers from the application web server). When using TMOS v12+ you may skip the creation of the HSTS iRule and create a custom
HTTP Profile with HSTS enabled instead. Attach either the iRule or the custom HTTP Profile (but not both!) to your HTTPS virtual server.
Name
Give the new iRule unique name. We use ir-add-HSTS-hdr.
Definition
Copy and paste the following iRule, omitting the line numbers:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
#
#
#
#
If site is TLS-protected, tell browsers to lock in
HTTPS to it for 31 days per rfc6797 (one warning per
connection is enough). A web server may supply its
own HSTS header to override this iRule.
when CLIENT_ACCEPTED priority 605 {
set add_HSTS_hdr [PROFILE::exists clientssl]
} ; #CLIENT_ACCEPTED 605
when HTTP_RESPONSE priority 605 {
if {$add_HSTS_hdr &&
![HTTP::header exists "Strict-Transport-Security"]} {
HTTP::header insert "Strict-Transport-Security" \
"max-age=2678400; includeSubDomains"
}
set add_HSTS_hdr false
} ; #HTTP_RESPONSE 605
DNS Records (Your DNS implementation)
Create suitable DNS records in your environment. In our example, the domain name for this service is app1.example.net, so we configure the following records. Add
appropriate reverse entries too, like 15.245.1.10.in-addr.arpa PTR app1.example.net.
DNS View
Name
Type
Value
public
app1.example.net
AAAA
2001:db8:16d:2::15
public
app1.example.net
A
192.0.2.15
internal
app1.example.net
AAAA
fdf5:f:5:c245::15
internal
app1.example.net
A
10.1.245.15
Profiles (Navigate to Local Traffic > Profiles)
Client SSL
(Profiles > SSL)
Name
Type a unique name. We use clientssl-app1
Parent Profile
clientssl
Certificate
Select the Certificate you imported onto the BIG-IP system for this deployment. We
select 2025-app1.example.net.
Key
Select the associated key. We select 2025-app1.example.net.
Chain
If your implementation requires an intermediate cert from a CA, select it from the list
Passphrase
Type the passphrase if the key is encrypted
If the web servers support multiple applications with TLS SNI, you can add additional certificates here
Server SSL
(Profiles-->SSL)
HTTP
(Profiles > Services)
Important: Only create this profile if
you are using BIG-IP v12.0 or later as
described in the iRules section
Name
Type a unique name. We use serverssl-app1
Parent Profile
serverssl
Server Certificate
require
Authenticate Name
app1.example.net
Trusted Certificate Authorities
Select the Trusted Certificate Authority you want to use. We select 2025-Private-CARoot-example.net
Name
Type a unique name (we use http-hsts-app1)
HTTP Strict Transport Security
Mode
Check the box to enable Mode. This enables the HSTS settings
Maximum Age
2678400
Include Subdomains
Ensure the Include Subdomains box is checked (Enabled)
Running the HTTP iApp template
The next task is to run the HTTP iApp template. In this example, you run the iApp template four separate times to create the virtual
servers for web traffic into DMZ1. To run the iApp template, click iApps > Application Services > Create. From the Template list,
select f5.http. Use the following table for guidance on configuring the iApp template for each scenario.
F5 Deployment Guide
38
Data Center Firewall
Question
1st: Answer IPv6 NS
2nd: Answer IPv4 NS
3rd: Answer IPv6 EW
4th: Answer IPv4 EW
Name
app1-public-6
app1-public-4
app1-intra-6
app1-intra-4
Which configuration mode
do you want to use?
Advanced – Configure
advanced options
Advanced – Configure
advanced options
Advanced – Configure
advanced options
Advanced – Configure
advanced options
Do you want to restrict client
traffic to specific VLANs?
Yes, enable traffic only on the
VLANs I specify
Yes, enable traffic only on the
VLANs I specify
Yes, enable traffic only on the
VLANs I specify
Yes, enable traffic only on the
VLANs I specify
On which VLANs should
traffic be enabled or
disabled?
Selected: /Common/external
(only!)
Options: (move all others here)
Selected: /Common/external
(only!)
Options: (move all others here)
Selected: /Common/internal
(only!)
Options: (move all others here)
Selected: /Common/internal
(only!)
Options: (move all others here)
How have you configured
routing on your web servers?
Servers have a route to clients
through the BIG-IP system
Servers do not have a route to
clients through the BIG-IP
Servers have a route to clients
through the BIG-IP system
Servers do not have a route to
clients through the BIG-IP
How many connections do
you expect to each web
server?
More than 64,000 concurrent
connections
More than 64,000 concurrent
connections
Create a new SNAT pool or
use an existing one?
Create a new SNAT pool
Create a new SNAT pool
What are the IP addresses
you want to use for the SNAT
pool?
2001:db8:16d:2::15
fdf5:f:5:c245::15
How should the BIG-IP
system handle SSL traffic?
Terminate SSL from clients,
re encrypt to servers (SSL
bridging)
Terminate SSL from clients,
re encrypt to servers (SSL
bridging)
Terminate SSL from clients,
re encrypt to servers (SSL
bridging)
Terminate SSL from clients,
re encrypt to servers (SSL
bridging)
Which Client SSL profile do
you want to use?
/Common/clientssl app1
/Common/clientssl app1
/Common/clientssl app1
/Common/clientssl app1
Which Server SSL profile do
you want to use?
/Common/serverssl app1
/Common/serverssl app1
/Common/serverssl app1
/Common/serverssl app1
Do you want to use BIGIP AFM to protect your
application?
/Common/accept-all 1
/Common/accept-all 1
/Common/accept-all 1
/Common/accept-all 1
How do you want to control
access to your application
from sources with a low
reputation score?
Allow all sources regardless
of reputation (because in this
configuration, we have a global
IP Intelligence policy)
Allow all sources regardless
of reputation (because in this
configuration, we have a global
IP Intelligence policy)
Allow all sources regardless
of reputation (because in this
configuration, we have a global
IP Intelligence policy)
Allow all sources regardless
of reputation (because in this
configuration, we have a global
IP Intelligence policy)
Would you like to stage a
policy for testing purposes?
/Common/app1 access
/Common/app1 access
/Common/app1 access
/Common/app1-access
Which logging profile would
you like to use?
afm-log-pfl
afm-log-pfl
afm-log-pfl
afm-log-pfl
What IP address do you want
to use for the virtual server?
2001:db8:16d:2::15
192.0.2.15
fdf5:f:5:c245::15
10.1.245.15
What FQDNs will clients use
to access the servers?
app1.example.net
app1.example.net
app1.example.net
app1.example.net
Do you want to redirect
inbound HTTP traffic to
HTTPS?
Redirect HTTP to HTTPS
Redirect HTTP to HTTPS
Redirect HTTP to HTTPS
Redirect HTTP to HTTPS
Which HTTP profile do you
want to use?
If using BIG-IP v12.0 or later (and not using the HSTS iRule) only:
Select the HTTP profile with
HSTS you created.
Select the HTTP profile with
HSTS you created.
Select the HTTP profile with
HSTS you created.
Select the HTTP profile with
HSTS you created.
Should the BIG-IP system
insert the X-Forwarded-For
header?
Do not insert X Forwarded For
HTTP header
Insert X Forwarded For HTTP
header 2
Do not insert X Forwarded For
HTTP header
Insert X Forwarded For HTTP
header 2
Do you want to create a new
pool or use an existing one?
Create a new pool
Select the pool created by
the first iApp, e.g.: app1public-6_pool
Select the pool created by
the first iApp, e.g.: app1public-6_pool
Select the pool created by
the first iApp, e.g.: app1public-6_pool
Which load balancing
method do you want to use?
Observed (member)
Which web servers should
be included in this pool?
How many seconds should
Slow Ramp time last?
1
2
3
/Common/webfront1 3
/Common/webfront2 3
60
This option chooses which policy to enforce. After you test and adjust policy ‘app1-access’ in staging mode you come back here to enforce it.
Needed with SNAT
Leave Port set to 443 and Connection Limit set to 0
F5 Deployment Guide
39
Data Center Firewall
Question
How many seconds should
pass between health
checks?
1st: Answer IPv6 NS
2nd: Answer IPv4 NS
3rd: Answer IPv6 EW
4th: Answer IPv4 EW
5
Which Web Acceleration
profile do you want to use
for caching?
Select the profile created by
the first iApp, e.g.: app1public-6_optimizedcaching
Select the profile created by
the 3rd iApp, e.g.: app1intra-6_optimized-caching
Which compression profile
do you want to use?
Select the profile created by
the first iApp, e.g.: app1public-6_wan-optimizedcompression
Select the profile created by
the 3rd iApp, e.g.: app1intra-6_wan-optimizedcompression
How do you want to optimize
client-side connections?
Select the profile created
by the first iApp, e.g.:
app1-public-6_tcp-wanoptimized
Which OneConnect profile
do you want to use?
Select the profile created by
the first iApp, e.g.: app1public-6_oneconnect
Select the profile created by
the 3rd iApp, e.g.: app1intra-6_oneconnect
How do you want to optimize
server-side connections?
Select the profile created by
the first iApp, e.g.: app1public-6_tcp-lan-optimized
Select the profile created by
the 3rd iApp, e.g.: app1intra-6_tcp-lan-optimized
Do you want to add any
custom iRules to this
configuration?
4
ir-add-HSTS-hdr 4
ir-add-HSTS-hdr 4
tcp-lan-optimized
ir-add-HSTS-hdr 4
Select the profile created by
the 3rd iApp, e.g.: app1intra-6_tcp-wan-optimized
ir-add-HSTS-hdr 4
If some portion of website is not served over HTTPS, or if the web server supplies its own HSTS header, you should omit this iRule.
Attaching the protocol security profile to the configuration produced by the iApp template
The next task is to attach the protocol security profile you created to the virtual servers created by the iApp template. This requires
first disabling the Strict Updates feature on each iApp Application Service. For any other changes you need to make to the
configuration, use the iApp template Reconfigure option.
To disable Strict Updates, click iApps > Application Services > app1-public-6 (or the name you gave the iApp). On the Menu bar,
click Properties. In the Strict Updates row, clear the checkbox to disable Strict Updates (you may have to select Advanced from
the Configuration menu first). Click Update. Repeat this process for each of the iApp application services you just created.
Navigate to Local Traffic > Virtual Servers > app1-public-6_vs (or name you gave the iApp followed by _vs). On the Menu bar,
click Security. From the Protocol Security list, select Enabled. From the Profile list, select the Protocol Security profile you created in
Creating a Protocol Security Profile on page 37 (afm-http-secy-pfl in our example), and then click Update. Repeat for each of
the virtual servers created by the iApp.
F5 Deployment Guide
40
Data Center Firewall
Additional information: Admitting Application Traffic to a DMZ
At the NS boundary of some application’s DMZ you typically apply both positive and negative security policy. The positive policy
is simple: you entertain only connections to the specified application (defined, in most cases, by LTM virtual servers on particular
addresses and ports). The negative policy is more nuanced: you prevent certain undesirable connections and exclude bad behavior,
but otherwise remain open to clients. The negative policy will have several layers. You might first use IP geolocation to exclude
connections from certain countries (in the example, countries on the ITAR list), then accept all other connections which don’t violate
some IP Intelligence and anti-DoS policies. (When you filter by IP geolocation consider layering on IP Intelligence policy to exclude
connections from origin-concealing proxies as the example here shows.) You should layer on a Protocol Security Profile to detect—
and stop—many attacks at the application protocol layer (e.g., HTTP). If you implement the BIG-IP ASM for comprehensive application
protection it will supersede the AFM Protocol Security Profile.
Of course you will apply a complementary (mainly positive) security policy at the DMZ’s EW trust boundary. The NS policy keeps
adversaries from, say, testing RPC exploits against your application servers. The EW policy keeps an adversary who finds, for
example, a command-injection vulnerability on a DMZ web server from using the compromised machine as a “jump server” to attack
the rest of the intranet.
Note that for destination addresses and ports in application virtual-server firewall rules you should put any:any rather than exact
values. That lets you update virtual-server addresses or ports later without having to modify firewall rules as well. This method does
not reduce security. Only packets which match a specific listener can ever reach the virtual-server context, so destination address
and port get checked automatically. (With a wildcard virtual server like mgmt-to-dmz-6-vs you will often use firewall rules to control
both source and destination addresses and ports. However, with a single-address-single-port application virtual server you need only
worry about source addresses & ports.)
The example shows some BIG-IP good practices. The App1 webservers accept HTTPS over IPv6 only, so you SNAT incoming
IPv4 connections and add X-Forwarded-For headers to demystify the webserver logs. You use a per-service SNAT pool rather than
SNAT AutoMap so even if you put more services in the DMZ you won’t risk port-exhaustion on your self IP. (It is customary to put the
application virtual-server’s own address in its SNAT pool to help identify traffic in network traces and server logs.) Finally, you create
separate virtual servers for traffic coming from the Internet and the intranet. That way you can use appropriate TCP and acceleration
profiles with each, and you can attach distinct security policies when appropriate (you might, for instance, attach different DoS
Protection Profiles to Internet-facing and intranet-facing virtual servers).
F5 Deployment Guide
41
Data Center Firewall
Securing a Branch-Office, Cloud-Infrastructure, or Business-Partner Link
Organizational networks often include VPN links to branch office, cloud infrastructure, or business partner networks which have
independent Internet connectivity. Typically such remote networks have firewalls, web proxies, and so on, but they still constitute
different trust environments. From a security policy point of view, such networks resemble DMZ’s—malicious actors might penetrate
their local NS defenses then exploit their EW connectivity to attack the intranet.
BIG-IP LTM includes support for a variety of VPN/tunnel schemes and protocols using IPSEC, TLS, etc. Whether you terminate a
site-to-site VPN on your BIG-IP device or route it through another device doesn’t really matter. Either way, the packets crossing it
come and go through BIG-IP VLAN/tunnel objects.
Setting up a VPN with LTM is beyond the scope of this document, but we look at configuring AFM to secure the link once you
establish it. If you do terminate a VPN on BIG-IP system you may have to add rules to the global-object network firewall policy and
possibly route-domain or virtual-server policies to admit VPN envelope packets. For instance, you might need to add a rule to policy
“control-from-Internet” to accept IPSEC traffic.
Because the trust boundary between a remote network and the intranet resembles that between a DMZ and the intranet, a similar
AFM configuration (for positive security policy) is appropriate. Note that this is an EW boundary; your data center firewall doesn’t
control the NS trust boundary between the Internet and the remote network. Typically you will trust a branch office more than a DMZ
and permit it to send more types of traffic to more of the intranet.
Use the following table for guidance on configuring the BIG-IP objects.
Address Lists (Navigate to Security > Network Firewall > Address Lists)
Name
all-BR-nets
Description
(example)
Notes (not a part of the configuration)
Addresses/Regions
Type or copy and paste the following addresses
into the Add new entry box:
10.8.0.0/16
fdf5:f:5:b000::/54
Name
br1-nets
Description
(example)
Addresses/Regions
Type or copy and paste the following addresses
Notes (not a part of the configuration)
into the Add new entry box:
10.8.1.0/24
All branch-office subnets (part of intranet-nets).
br1 = branch office #1
fdf5:f:5:b001::/64
Rule Lists (Navigate to Security > Network Firewall > Rule Lists. See previous Rule List sections for specific guidance)
Rule List Name
Rule Name
Protocol
Source address
Src. Port
Src. VLAN
Dst. Address
Dst. Port
Action
Accept
ok-br1-udp
UDP
br1-nets
Any
vpn-BR1
intranet-nets
udp-EW-basic
udp-EW-generic
ok-br1-tcp
TCP
br1-nets
Any
vpn-BR1
intranet-nets
tcp-EW-basic
tcp-EW-generic
tcp-rdp-smb
Accept
ok-to-br1
TCP
intranet-nets
Any
Any
br1-nets
tcp-rdp-smb
Accept
Accept
Accept
permit-br1-to-intra
permit-intra-to-br1
ok-tcp-mgmt
TCP
net-mgmt-nets
Any
Any
all-BR-nets
tcp-EW-generic
tcp-rdp-smb
ok-udp-mgmt
UDP
net-mgmt-nets
Any
Any
all-BR-nets
snmp-mgr
udp-dhcp
permit-mgmt-to-BR
Network Firewall Policies (Navigate to Security > Network Firewall > Policies)
Policy notes
BIG-IP AFM may detect an overlap between rules in policy “intra-to-br1” permitting tcp-rdp-smb traffic to br1 from both the
management network and the intranet generally. This is tolerable because removing the redundancy would make it harder
to crack down later on non-management use of tcp-rdp-smb—you would have to edit two rule lists instead of just one.
Policy Name
intra-to-br1
Rules
Click the Add button and add the following rule.
F5 Deployment Guide
Click Finished and then from the Policies table, click the Policy you just created.
Name
mgmt2BR
Type
Rule List
Rule List
permit-mgmt-to-BR Click Repeat and then add the following rule.
Name
intra2br1
Type
Rule List
Rule List
permit-intra-to-br1 Click Finished.
42
Data Center Firewall
Click Finished and then from the Policies table, click the Policy you just created.
Policy Name
br1-to-intra
Rules
Click the Add button and add the following rule.
Name
br1-intra
Type
Rule List
Rule List
permit-br1-to-intra Click Repeat and then add the following rule.
Virtual Servers (Navigate to Local Traffic > Virtual Servers)
IPv4 intranet to branch office virtual server
Name
intra-to-br1-4-vs
Type
Forwarding (IP)
Destination Address
10.8.1.0/24
Service Port
* (*All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the internal VLAN you created.
Source Address Translation
None
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
Next, we stage the initial policies. After you generate some test traffic and make any adjustments suggested by log
analysis you can move your policies from staging to enforcement.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network Firewall
Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the intranet to branch office policy you just created (intra-tobr1 in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
IPv6 intranet to branch office virtual server
Name
intra-to-br1-6-vs
Type
Forwarding (IP)
Destination Address
fdf5:f:5:b001::/64
Service Port
* (*All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the internal VLAN you created.
Source Address Translation
None
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network Firewall
Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the intranet to branch office policy you just created (intra-tobr1 in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
IPv4 branch office to intranet A virtual server
Name
br1-to-intraA-4-vs
Type
Forwarding (IP)
Destination Address
10.0.0.0/8
Service Port
* (*All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the branch office VLAN you created (vpn-BR1 in our example).
Source Address Translation
None
F5 Deployment Guide
43
Data Center Firewall
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network Firewall
Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the branch office to intranet policy you just created (br1-tointra in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
IPv4 branch office to intranet C virtual server
Name
br1-to-intraC-4-vs
Type
Forwarding (IP)
Destination Address
192.168.0.0/16
Service Port
* (*All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the branch office VLAN you created (vpn-BR1 in our example).
Source Address Translation
None
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network Firewall
Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the branch office to intranet policy you just created (br1-tointra in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
IPv6 branch office to intranet virtual server
Name
br1-to-intra-6-vs
Type
Forwarding (IP)
Destination Address
fdf5:f:5::/48
Service Port
* (*All ports)
Protocol
* All Protocols
Protocol Profile (Client)
Select the TCP Loose Fast L4 profile you created (tcp-loose-fastl4 in our example).
VLAN and Tunnel Traffic
Select Enabled On, and then select only the branch office VLAN you created (vpn-BR1 in our example).
Source Address Translation
None
Click Finished and then from the Virtual Servers table, click the virtual server you just created.
On the Menu Bar, click Security > Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network Firewall
Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Enabled. From the Policy list that appears, select the branch office to intranet policy you just created (br1-tointra in our example).
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
Select Enabled. From the Available list, select the Log profile you created in Configuring AFM High-Speed Logging on
page 23 (afm-log-pfl in our example).
F5 Deployment Guide
44
Data Center Firewall
Additional Information: Securing Links to Remote Networks
There are two additional concerns when dealing with a cloud infrastructure or business partner network instead of a branch office: IP
addressing and narrow service design.
A remote network may use addresses you did not assign. If they conflict with yours (a common occurrence when both sides use IPv4
private addresses) then address translation, preferably BIG-IP SNAT, is required. If they don’t (e.g., both sides use IPv6 ULA prefixes),
either routing or (S)NAT will be needed. Routing is preferable to address translation but watch out for a business partner who runs
public IPv6 addresses on his intranet—to avoid problems with his firewall policies you may have to route to just his internal subnets via
the VPN and continue to reach the partner's external services via the Internet.
Many VPN’s to cloud infrastructure or business partner networks support just one or a few tightly-constrained services like EDI or
batch file transfer. That is a favorable situation: just create specific virtual servers to handle those connections. That is more secure
(and easier to audit) than IP forwarding virtual servers gated by AFM network firewall policies.
You may have to create some application-specific virtual-servers in any case to support ALG’s for protocols like FTP or SIP.
The final issue is how much to restrict connection attempts from the intranet to the remote network. If you don’t expect too much
configuration churn, it is good practice to filter outbound connections like inbound ones—to specific addresses and ports. If nothing
else, that may hobble APT’s on the intranet trying to spread to the remote network.
Note that you might choose to accelerate branch office traffic with F5 BIG-IP Application Acceleration Manager (AAM). Configuring
AAM is beyond the scope of this document, but it does not obviate any network security principles.
The example describes a branch office link. Branch office workers are given fairly free connectivity to standard applications in the
intranet. Among enhancements you should consider: limiting destination networks (for example, to the data center only, though that
would prevent RDP from branch office workstations to PC’s in headquarters offices); expanding destination ports to permit additional
applications (for example, additional VDI services); and channeling web traffic to a filtering proxy in the intranet so it can scan for
malware and/or data exfiltration. For instance, you could create virtual servers listening to TCP ports 80 and 443 on the VPN VLAN/
tunnel to send all web traffic directed to the intranet from the branch office transparently through the proxy.
The example assumes there are no business servers in the branch office—just PCs someone might wish to access remotely—so
access to the branch office network is quite limited for most intranet users. However, the management network is privileged as usual
to facilitate remote administration of equipment in the branch office. The example firewall policy blocks web traffic from the intranet
to the branch office to prevent intranet users routing through the branch office to avoid the home office web proxy (If you create a rule
to allow only the main web proxy to send web traffic to branch offices you can put web servers there while ensuring that head office
users send all traffic through the main proxy—even traffic to servers in branch offices). If necessary you could add rules to permit
access to specific services on the remote network.
F5 Deployment Guide
45
Data Center Firewall
Defending the BIG-IP system Itself with AFM
The details of security management for BIG-IP devices are mostly beyond the scope of this document. That is not to minimize their
importance. Network infrastructure devices, especially those enforcing security policy are prime targets for malicious users. BIGIP platform security guidance is collected at F5 SOL13092: Overview of securing access to the BIG-IP system. F5 also offers a
deployment guide and iApp to help you secure BIG-IP devices according to NIST Special Publication 800-53r4, Recommended
Security Controls for Federal Information Systems, available at http://f5.com/pdf/deployment-guides/nist-sp-800-53-r4-dg.pdf.
However, since not all BIG-IP devices have AFM provisioned, neither SOL13092 nor the F5 NIST SP800-53 deployment guide
explains the role of AFM in BIG-IP platform security.
This deployment guide does show how to protect the BIG-IP management (network) port and self IPs using AFM network firewall
policies.
Protecting the BIG-IP Management Port with AFM
It is wise (though not always feasible) to separate network management traffic from ordinary application traffic onto special
"management network" subnets and VLANs. The BIG-IP system is optimized for such an environment (as the BIG IP management
port must be homed on a different subnet than any TMM self IP). Some organizations, though, allow management traffic to originate
from non-management network subnets. Such traffic, most commonly from administrators using workstations on corporate office
subnets is merely routed through the management network.
The BIG-IP management port is pretty well hardened. To protect it further using a network firewall policy you must be able to
recognize legitimate management traffic by source address. That is fairly easy when all such traffic comes from management network
subnets. Otherwise you may have to maintain lists of other trusted subnets, use AFM identity-based security, or use an AAA gateway
like BIG-IP APM to SNAT or VPN authorized traffic from less trusted networks onto the management network.
The model network architecture has a management network which owns all the subnets of 192.168.192.0/18. You can use AFM
to ensure that only traffic from those subnets reaches the BIG-IP management port. You must update the nameless firewall policy
permanently enforced on the management port to accept packets from management subnets and explicitly Drop packets from all
other sources. (By default, packets which do not match any rule in the management-port context are Accepted. There is no fallback
to the final (global-default) context from the management-port context.)
!
Warning It is possible to lock yourself out of a BIG-IP management port by making an error in the management port firewall
policy. Then, if you cannot access TMSH or the BIG-IP management GUI through some other port you may not be
able to correct the error.
Before you configure any AFM firewall policy on the management port, ensure you have access to:
• The combination of (1) a BIG-IP self IP address with Port Lockdown set to “Allow Default” and (2) a permissive
firewall policy (such as /Common/allow-all) enforced on that self IP object (allowing access from your PC to
ports 22 and 443 at least) and (3) System / Platform / User Administration / SSH Access set to “Enabled”; or
• The serial console port on a hardware BIG-IP device; or
• The KVM virtual console of a BIG-IP VE device.
Test your alternate access before changing the management port firewall policy. After you have configured and
successfully tested your management-port firewall policy, you may adjust the firewall policy and/or Port Lockdown
setting on any self IP you used as an alternate management access point.
Securing the management port
On each BIG-IP (big-s1 and big-s2 in our example—management-port settings are not synchronized across devices), to secure the
management port add the following firewall rules in order.
Management Port Rules (Navigate to System > Platform > Security > Add Rule)
Name
ok-mgmt-nets
Protocol
Any
Source > Address/Region
Add net-mgmt-nets and fe80::/10
Source > Port
Any
Destination > Address/Region
Any
Destination > Port
Add udp-EW-basic and udp-EW-generic
Action
Accept
Logging
Disabled
F5 Deployment Guide
Click Repeat and then add the following rule
46
Data Center Firewall
Management Port Rules (Navigate to System > Platform > Security > Add Rule)
Name
no-others
Protocol
Any
Source > Address/Region
Any
Source > Port
Any
Destination > Address/Region
Any
Destination > Port
Any
Action
Drop
Logging
Enabled
This completes the management port rules.
F5 Deployment Guide
47
Data Center Firewall
Deep Packet Inspection with AFM (Blocking Teredo)
BIG-IP AFM is not a dedicated deep-packet-inspection tool like an IDS. However, one job of a data center firewall is to detect and
block attempts to evade network security policy. That is a different project than looking for malware signatures in application data.
Some evasion techniques can only be recognized by fairly sophisticated deep-packet-inspection filters. The BIG-IP platform lets you
implement complex filters using iRules. For example, you may use an iRule to block Teredo.
Teredo (rfc4380) is a protocol for conveying IPv6 packets across the IPv4 Internet. It is designed to "worm" through NATs and stateful
firewalls. A Teredo tunnel is a sort of VPN; though unencrypted, it conceals the source, destination, and content of IP traffic it carries
from most of the firewalls through which it passes. Teredo is a legitimate network protocol which may be misused to evade network
security policy in many situations.
All versions of Microsoft Windows from XP-SP2 onward incorporate Teredo client drivers. Drivers also exist for other operating systems
such as Linux and Mac. Special-purpose platforms may also use Teredo; Microsoft Xbox, for instance, routinely creates Teredo tunnels.
Because Teredo may be used to smuggle all sorts of network traffic past security policy (including APT traffic, botnet control channels,
AUP-violating web traffic, etc.), it is generally wise to block it.
However, it is not easy to block Teredo tunnels without interfering with other, sanctioned network traffic. A couple of UDP ports are
well-known to be used by Teredo but they are neither reserved for Teredo nor necessary for it to work. Closing the well-known ports
won't prevent the use of Teredo but will disrupt other application traffic. The only sure way to block Teredo with a typical network
firewall rule is to close all UDP ports—a cure commonly considered worse than the disease.
You may use a BIG-IP AFM iRule to block Teredo tunnels on all ports without disrupting other application traffic. The iRule analyzes
the initial packets of UDP flows to detect Teredo connection attempts and reject them. This is particularly efficient on the BIG-IP
system because only one packet per flow need be examined. Attached to the virtual server for outbound UDP traffic the iRule
prevents use of Teredo tunnels from the intranet.
Create the following iRule then attach it to virtual server for outbound UDP traffic (udp-icmp-outbound-4-vs in our example).
iRules (Navigate to Local Traffic > iRules)
Name
Give the unique name. We use ir-block-teredo
Definition
Copy and paste the following iRule, omitting the line numbers:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
F5 Deployment Guide
when FLOW_INIT priority 400 {
# to prevent Teredo tunnels we recognize and block Router
# Solicitation messages to Teredo servers. This closes all
# ports to Teredo without interfering with other services
# on those ports. (Teredo is not limited to port 3544; for
# example, Microsoft Xbox uses 3074 for Teredo.)
#
# (Attach this iRule to your forwarding-virtual-server for
# outbound UDP traffic.)
if {([IP::protocol] != 17) ||
([set len [eval {UDP::payload length}] < 48)} { return }
# Teredo authentication header may precede RS
set ofs 0
binary scan [eval "UDP::payload 4"] Scc aID CL AL
if {$aID == 1} {
set ofs [expr {13 + ($CL & 0xff) + ($AL & 0xff)}]
if {$len < (48 + $ofs)} { return }
}
#extract possible Teredo RS
set buf [string range [eval "UDP::payload [expr {$ofs + 48}]"] $ofs end]
#Teredo RS will have a bunch of fields set to particular values,
#we look for IPv6 version, length, next header, destination
#address upper and lower portions, source address prefix and
#flag bits, and ICMPv6 RS type-code. Those seem sufficient but
#if we get false positives we can also check the hop-count,
#reserved-word contents (~0), and/or validate ICMPv6 checksum.
binary scan $buf [email protected]@24wWsSI \
Iver Ilen Inxt Ihop Ipfx Iflg Id1 Id2 \
RStypc RSchek RSrsvd
if { ($Iver eq "6") && ($RStypc == 133) &&
($Id1 == 0x2ff) && ($Id2 == 2) &&
($Ilen == 8) && ($Inxt == 58) &&
($Ipfx == 0x80fe) && (($Iflg & 0x7fff) == 0) } {
#this packet is almost certainly a Teredo RS
reject ; #no tunnel for you!
}
#otherwise this packet is not a Teredo RS, let it pass
} ; #FLOW_INIT 400
48
Data Center Firewall
Appendix A: Supplemental Information
In this appendix, you'll find additional and complementary information to the guidance provided in this document.
Dual-Stack IPv6/IPv4 Networking
The world ran out of IPv4 addresses in 2014. Many firms and most individuals haven’t quite noticed but now IPv6 is the only platform
for growth. Internet service providers and their largest customers have already made the change. Every organization must follow suit.
A decade of industry experience with IPv6 has taught everyone that a dual-stack network with intelligent proxy gateways is key to a
successful transition from IPv4 to IPv6. Though dual-stack clients are handy, it is not critical for servers to be dual-stacked so long as
the network is and gateways are provided to help application traffic get from one addressing realm to the other.
By definition, a dual-stack network carries both IPv4 and IPv6 traffic—so it can potentially carry harmful traffic in either realm. Even if
your network is not yet officially dual-stack, you have IPv6 traversing the network already because every version of Microsoft Windows
since XP has been dual-stack. At this time, the real problem with dual-stack networking is maintaining your security posture. You
have many adversaries eager to use IPv6 to evade IPv4-based controls and vice-versa. Your team may be less experienced with
IPv6 vulnerabilities. The question is whether your dual-stack network will be efficient and secure. The only way to ensure either is to
configure your infrastructure so both IPv4 and IPv6 are fully managed. You must make security planning, policies, enforcement, and
monitoring coextensive for IPv6 and IPv4—ideally using tools like the BIG-IP system which unify configuration and processing of both.
The era of tunneling IPv6 around is over for businesses and other organizations (for end-users that is—carriers and ISP’s have special
uses for tunneling and cross-realm NAT schemes). Business-class ISP’s are dual-stack; they give simultaneous IPv4 and IPv6 service.
You want both, so you can grow using IPv6 addresses while interoperating with the IPv4 installed base. Getting rid of tunnels gets rid
of many operations and security challenges as well.
Of course, IPv6 clients on a dual-stack network can talk to IPv6 servers and IPv4 clients can talk to IPv4 servers. In the past it was not
obvious how clients in one realm should talk to servers in the other. Given a large installed base of IPv4-only servers and a growing
supply of IPv6-only servers, the industry experimented with various NAT schemes to facilitate cross-realm communication. None was
really satisfactory. However, dual-stacked proxy gateways worked beautifully, and the F5 BIG-IP is the archetypical proxy gateway.
BIG-IP virtual servers can accept traffic on IPv4 and/or IPv6 addresses, and then proxy (and load balance) application traffic to
destinations in the other realm—or the same realm; the client neither knows nor cares. You may add a server using IPv6 alongside
one using IPv4 and then use the BIG-IP system to serve even single-stack clients from both servers at the same time.
In the optimal dual-stack network design—experimentally refined since the late-2000’s—you home your real servers logically behind
the BIG-IP device (where it doesn’t matter which stack they run) and publish both IPv6 and IPv4 virtual servers to the intranet and
to the world. Clients connect via the address realm they find most convenient and the BIG-IP system proxies the connections.
When your single-stack clients need to connect to some remote service which is only offered in the other realm, you may proxy the
outbound connection—the distant server never knows it is (logically) behind your BIG-IP device. The BIG-IP system even supports
dynamic gateway configuration with automatic DNS translation for IPv6 clients.
The model network architecture for this Deployment Guide is dual stack and so are the practical examples. For instance, the network
firewall policy examples filter IPv4 and IPv6 traffic simultaneously to demonstrate unified security policy in the dual-stack network.
BIG-IP in a Dual-Stack Network
This Deployment Guide shows how to configure your BIG-IP for dual-stack networking. However, there are a few points which the
examples don’t illuminate:
hh Y
ou should not add any server to a particular LTM pool (as a member) more than once, even if that server has both IPv6
and IPv4 addresses (i.e., is multihomed).
hh Don’t attach a Fast HTTP Profile to a virtual server with an IPv6 address (see SOL11449).
hh Y
ou may configure dynamic IPv4 to IPv6 DNS translation and set up dynamic UDP and TCP gateways to allow IPv6-only
clients to connect to IPv4-only services without identifying those services in advance. The details are beyond the scope of
this document, but not very difficult.
hh O
n very busy IPv6 networks, AFM DoS Protection Device Configuration may classify benign ICMPv6 messages with type
numbers above 132 as DoS attacks. You may adjust the detection threshold if necessary.
F5 Deployment Guide
49
Data Center Firewall
BIG-IP AFM Overview
As a reminder, this document is based on BIG-IP/TMOS version 11.6 and 12.0. Subsequent releases of AFM may include features
beyond those discussed here.
The major components of AFM are Network Firewall, IP Intelligence, Denial-of-Service (DoS) Protection, and Protocol Security. All are
integrated into the BIG-IP full-proxy architecture and with LTM especially.
Only the Protocol Security component normally inspects packets very deeply, but the Network Firewall may perform deep packet
inspection with iRules (for a practical example, see Deep Packet Inspection with AFM (Blocking Teredo) on page 48), and the
DoS Protection DNS and SIP Profiles deal with complex protocols. Of course AFM is just one part of the BIG-IP Application Delivery
Firewall. The BIG-IP system enforces protocol compliance at multiple layers using LTM configuration objects and ALG’s.
You typically use the Network Firewall to apply some negative and much positive security policy. The limitation of the Network
Firewall is that it gets only one chance to make a decision about each connection, chiefly by looking at the L2–4 characteristics of
its initial packet. That really is sufficient control for 99+% of traffic across the network, but in nearly all customer environments a few
superficially-legitimate connections will actually come from malicious actors who abuse connections to transmit attacks.
IP Intelligence, DoS protection, and Protocol Security support mostly negative security policy. IP Intelligence lets you recognize and
exclude connection requests from known-bad origins. DoS Protection and Protocol Security recognize and limit misuse of active
connections which seemed legitimate when initiated.
AFM Protocol Security defeats many attacks at the HTTP or DNS protocol level. However, to defend against attacks on application
logic (for example, web SQL-injection attacks, which are transmitted in syntactically-valid HTTP requests) you will want an applicationlevel firewall like F5 BIG-IP Application Security Manager (ASM).
AFM’s tight integration with other BIG-IP features is essential to overall system capability and performance. For example, BIG-IP LTM
TLS (SSL) processing enables Protocol Security to protect secure websites.
The examples in this document illustrate at least basic use of each major AFM component.
From Zone-Based Firewall to Application Delivery Firewall
With the BIG-IP system, the constructs interface and zone commonly used to configure less advanced firewalls are superseded by
listener and application. Application-based security policy is both more granular and better-informed than zone-based policy. Tying
policy to applications frees administrators to move application-related resources around without the traditional problem of detaching
them from their carefully-crafted security controls. This helps avert the gradual lessening of security due to “rule rot” and the
accumulation of abandoned zone-policy rules which degrade firewall performance and auditability.
See the F5 Tech Brief Replacing Abstract Zones with Real Application Security Policy for a high-level discussion.
F5 Deployment Guide
50
Data Center Firewall
Network Firewall
This section contains additional information on the AFM Network Firewall.
AFM Network Firewall Structure
An AFM network firewall policy is an ordered collection of firewall rules attached to a BIG-IP traffic-handling object (such as a listener
like a virtual server) which decides whether that object will process an incoming packet. The packet (more precisely, a set of attributes
describing the packet—such as its IP protocol, source and destination addresses and ports, the VLAN from which it arrived, etc) is
tested against the policy’s rules, in order, until a match is found or no rules remain. The action of the first matching rule determines the
result of the policy. (If no rule matches the packet another policy will judge it. A last-resort policy discards never-matched packets.)
Each object belongs to a firewall context. All similar objects (e.g., virtual servers) belong to the same context. Each context belongs
to a hierarchy that determines the order in which policies are applied to each packet entering the BIG-IP system.
The contexts are: global; route-domain; virtual-server; self IP; global-default (which is more easily understood as final); and
management-port. The primary context hierarchy includes (in order of policy application) the global, route-domain, virtual-server –or–
self IP, and final (global-default) contexts (see Figure 2). The second hierarchy includes only the management-port context.
The context hierarchy helps keep policies simple. The global-context policy, for example, may contain general rules (like one to
discard packets improperly sourced from multicast addresses) which you would otherwise have to tediously place into every virtualserver’s policy. There is no “VLAN context,” but VLAN-specific rules in the global context give almost the same effect.
You may attach up to two policies to most objects (the management port is an exception): one enforced policy and one staged.
The action determined by an enforced policy actually controls whether a packet proceeds through the BIG-IP system. The action
determined by a staged policy is logged but not effectuated. The point of staging a policy is to find out how it would affect traffic if
you were to enforce it—so you may debug policies before applying them. Staged policies have an auxiliary use: counting and logging
selected packets without changing the mainline configuration of the BIG-IP system. For example, you could count connections
initiated from subnet 203.0.113.0/24 by staging a policy with a rule of the form “Accept src 203.0.113.0/24:any dst any:any.” Staged
rules don’t interfere with traffic, but they do count matches for your review.
The global and management-port contexts each contain only one object. The management- port object has one nameless but
editable policy permanently enforced on it.
When you have not enforced a named policy on an object, BIG-IP AFM enforces a default implicit policy which matches packets from
any source addressed to the object according to its normal properties. For example, the implicit policy for a virtual server listening on
VLAN “external” for web traffic to 203.0.113.7 would amount to “take action X on packets from source= any:any, via VLAN= external,
protocol= TCP, to destination= 203.0.113.7:80.” All implicit policies on virtual servers and self IPs take the same action X, which you
set in Security > Options > Network Firewall > Virtual Server & Self IP Contexts to one of Accept, Reject, or Drop.
When you set the action for implicit policies on virtual servers and self IP’s to Accept, AFM runs in ADC mode. That means
BIG-IP traffic-handling objects process the packets you would expect them to and all other packets get discarded. Implicit policies
are dynamic. They are neither visible in per-object configuration nor recorded in BIG-IP configuration files. However, you may see the
implicit policies active at run time by navigating to Security > Network Firewall > Active Rules and selecting Context “All (Show implied
rules)”. Since global and route-domain objects are promiscuous, the implicit policies for them match all packets (those packets are
subsequently filtered by policies further down the context hierarchy).
Some organizations wish all firewall policy to be explicit and visible in configuration files—to facilitating auditing, or to take a belt-andsuspenders approach to configuration change control. You may run AFM in Firewall mode by setting the action for implicit policies
to Reject or Drop. In Firewall mode, no virtual server or self IP object will process any traffic until you enforce on it a named firewall
policy which has at least one rule with action “Accept”. To activate a new service you must, e.g., both create a virtual server and
enforce a named policy on it. Firewall mode is not inherently more secure than ADC mode (though it is more cumbersome). In both
modes network firewall policy controls all packets which arrive to the BIG-IP. (Per the model network architecture, the examples in this
document show AFM in Firewall mode.)
F5 Deployment Guide
51
Data Center Firewall
Figure 3: The Firewall Policy Context Hierarchy (omits management-port context)
The firewall policy on the global object—the sole global-context policy—applies to all packets reaching the BIG-IP except those
arriving to the management port (which has an independent context and policy). If a packet traverses the global context it will reach
the route-domain context where AFM will apply the policy enforced on the relevant route-domain object. If the packet traverses the
route-domain context it must find a listener (either a virtual server or a self IP object) or it will be discarded. Assuming the packet finds
a listener, then in most cases AFM will apply the policy enforced on that object (in the virtual-server or self IP context as appropriate).
For performance reasons, a firewall rule in the global or route-domain context may authorize a packet to bypass the virtual-server- or
self IP-context policy (for instance, when a packet arrives from a highly-trusted subnet).
A typical incoming packet traverses the global and route-domain contexts, then the virtual-server or self IP context. In every context,
the packet is tested against the applicable policy’s rules in order. The first rule to match a packet determines whether it will be
forwarded to the next context (or otherwise processed). When the first matching rule has action Reject or Drop, AFM discards
the packet (silently for Drop, or with notice to the sender for Reject). However, in the global and route-domain contexts, if the first
matching rule has action Accept, or if no rule matches the packet, then the packet will merely be forwarded to the next context. In
other words, actions Drop and Reject are always final in any context, but action Accept is provisional until a packet reaches either the
virtual-server or self IP context, where Accept leads to processing by the listener.
It may happen that no rule in a policy matches a packet. In that case the packet is forwarded to the next context. However, the next
context after the virtual-server and self IP contexts is the final (global-default) context where a fixed single-rule policy matches and
discards all packets that reach it. In other words, to get processed by a listener, a packet must match an Accept rule in the virtualserver or self IP context, or an Accept-Decisively rule in the global or route-domain context. Accordingly, in the virtual-server and
self IP contexts, actions “Accept” and “Accept Decisively” are equivalent.
Three invisible rules are always logically prepended to the global-context policy, where they may match a packet before any normal
rule gets a chance to match it. The first invisible rule is the most important: it Accepts Decisively any packet which belongs to an
established flow to or through the BIG-IP.
A flow represents a network connection, i.e., a bidirectional stream of closely-related packets. For example, when a virtual server
replies to a TCP SYN packet from a client, it creates a flow to recognize additional packets in the new TCP connection by the classic
five-tuple of {protocol, source address and port, destination address and port} plus some additional criteria such as route domain
and VLAN. Outbound connections create flows too. Flows are logical entities represented by entries in a flow table (you can use the
F5 Deployment Guide
52
Data Center Firewall
TMSH command show /sys connection to inspect the flow table). Since the BIG-IP system operates most of the time as a full
proxy, what seems to be a single (logical) connection at the application layer commonly has two flows, one each on the “client” and
“server” sides of the BIG-IP device.
In combination with the first invisible rule, flow-table entries act as ephemeral firewall rules to permit packets for established
connections to pass the firewall very efficiently. AFM’s support for flows also removes any need for “reverse” firewall rules. Nearly
every packet that leaves the BIG-IP is part of a flow (including packets sent by BIG-IP processes like service heath monitors), so AFM
will recognize related packets coming the other way and Accept them Decisively in the global context. (A packet accepted by AFM
may still be discarded for some other reason.)
The first invisible rule means that once the initial packet of a new connection has been accepted for processing by a virtual-server or
self IP object, other packets (flows) associated with that connection are accepted by association. (Of course when a traffic-handling
object like a virtual-server refuses or terminates a connection, that removes all associated flows.)
The second invisible rule discards every TCP packet which does not belong to a flow except for SYN packets requesting new
connections.
The third invisible rule Accepts Decisively any packet for which an ALG (Application Layer Gateway) is specifically listening. Like other
listeners, ALG’s may accept connection requests and establish new flows. This rule has the same security implications as the first
invisible rule because ALG listeners and connections are always auxiliary to ordinary listeners’ connections.
The three invisible rules make AFM a stateful firewall so far as packet-matching rules are concerned. Of course, the BIG-IP is really an
Application Delivery Firewall. BIG-IP is much more “stateful” than most network firewalls because BIG-IP maintains connection state
at Layers 2–7 and enforces protocol compliance at every layer.
Finding a Listener
To choose which policy will judge a packet in the route-domain, virtual-server, or self IP context, AFM interacts with a fundamental and
critical feature of BIG-IP: each (accepted) connection will be handled by the listener (virtual server, usually) that offers the most specific
match for it based on VLAN, IP address, port, and protocol.
When defining firewall rules you must order them carefully to make “more specific” ones match before “less specific.” For example,
to let traffic reach one host on a subnet while blocking access to all other hosts on that subnet you would put rules like “no-net1:
deny src any dst 203.0.113.0/24” and “ok-host7: permit src any dst 203.0.113.7/32” in the order (ok-host7, no-net1). If you were to
reverse that order, the “block whole subnet” rule (no-net1) would match before the “allow particular host” rule (ok-host7) so no packet
would ever reach 203.0.113.7.
Matching connections to BIG-IP listeners doesn’t work that way.
The BIG-IP automatically chooses the best-matching listener for each connection. Suppose you configure listeners P on
203.0.113.0/24:any and Q on 203.0.113.7/32:any. When a packet addressed to 203.0.113.7:993 arrives, BIG-IP will deliver it to the
more-specific listener Q. A packet addressed to 203.0.113.5 would go to the less-specific listener P. Suppose you replace Q with
two listeners R on 203.0.113.7/32:80 and S on 203.0.113.7/32:22. In that case a packet addressed to 203.0.113.7:993 would find
listener P on 203.0.113.0:any. Because it offers the best match—the others are too picky.
This approach lets BIG-IP listeners “peel off” portions of the traffic stream for various kinds of treatment. For example, each ALG you
configure has a listener which attracts precisely the traffic the ALG is meant to handle (based on IP protocol, port, etc.) while other
traffic goes to a more generic listener (or to the bit-bucket).
Virtual servers are the most important type of listener. Virtual servers provide the nexus between application traffic and security policy
(really, all kinds of application-delivery policy) at all layers. You bind network firewall policies to applications as well as mere routing
paths (e.g., IP forwarding virtual servers) by applying them in the virtual-server context.
AFM with Virtual Servers for Positive Security
BIG-IP’s listener-matching model embodies a positive security policy. Indeed, to create a pretty effective firewall you can simply define
listeners for desirable traffic and let BIG-IP discard everything else. However, the generality of BIG-IP with AFM lets you choose from
two approaches to enforcing strict security policy at trust boundaries. You may configure specific virtual servers to collect and proxy
each class of traffic from one trust environment to selected resources in another, or you may configure IP forwarding virtual servers
and use AFM network firewall policy to filter packets going through them. The second approach resembles a traditional interface-andport (zone-based, router-type) firewall. It is simple to set up but tricky to maintain. The first (specific-virtual-server) approach is a bit
more trouble to set up but easier to “keep tight.”
F5 Deployment Guide
53
Data Center Firewall
In practice you will use both approaches, as shown in the examples here. When you want to enforce security above Layer 4 (for
instance, using an AFM Protocol Security Profile) and when you wish to be very careful (for instance, when exposing a service to the
Internet) you will configure application-specific virtual servers backed by specific resource pools. When packet-based security will
cover most of your needs (for example, controlling EW traffic to well-known application ports) and when you anticipate that network
users (say, sys-admins in the data center) may move services and clients without telling you, you will set up generic virtual servers and
use network-firewall policy to keep traffic through them fairly clean.
(Perhaps the toughest reorientation for experienced zone-firewall admins mastering BIG-IP is to stop trying to guess which interface
an incoming packet is likely to leave through. Let the BIG-IP listener worry about that. BIG-IP is more than a router, and in many
cases an incoming packet will never leave because it is part of a flow to one side of a BIG-IP virtual server (proxy) that will forward
only payload data (possibly transformed) to the remote destination. The critical security question is: should an incoming connection
request be accepted (and corresponding flows established)? The combination of an appropriate listener and the AFM policies or
profiles attached to it will provide the answer. Once a connection request is vetted, LTM will handle communications to the ultimate
resources.)
AFM Network Firewall Rules
Network firewall rule lists collect rules for use in one or more network firewall policies. Each rule in a policy may match packets directly
(by protocol, addresses, etc.) or refer to a rule list. Rules in rule lists must match packets directly; rule lists cannot be nested. When
a policy rule refers to a rule list, packets are tested against the rules in that list in order, as if those rules appeared in the policy at the
point of reference.
For example, here are two rules that make up a policy to control access to an email gateway:
Policy Name
email-gw
Rule Name
Protocol
Source address
Src. Port
Src. VLAN
Dst. Address
Dst. Port
Action
no-419s
TCP
Nigeria (NG)
Any
external
Any
25
Drop
ok-smtp
TCP
Any
Any
external
Any
25
Accept
If you modify a rule list—or a rule inside a rule list—all policies which refer to that rule list will be changed. However, altering a directmatching rule in one policy will not affect other policies even if they include similar rules. Since one rule list cannot reference another,
active rules are never more than one step removed from a policy.
Even though you give each rule inside a policy or rule list a name, you cannot refer to that rule elsewhere. To use a particular rule in
more than one policy you must add it to a rule list (which you can reference in multiple policies).
Address lists collect addresses for use in one or more rules. Address lists may be nested. The order of addresses in a list is not
significant.
Each address in a rule or an address list is static, dynamic, or symbolic. Static addresses consist of an IPv6 or IPv4 prefix and
(subnet-) mask length (CIDR notation). If the mask length is 128 (IPv6) or 32 (IPv4) the address specifies a host (when you don’t
supply a mask length, a length of 128 or 32 is presumed). Otherwise the address specifies a subnet or a block of adjacent subnets.
A static address in a rule matches a packet address when the leftmost mask-length bits of both addresses are the same. For
instance, a subnet address matches all host addresses in that subnet. Dynamic addresses include geolocation and user-identity
indicators. At run time, rules which refer to dynamic addresses test whether some actual (source or destination) address in a packet
matches a given geolocation indicator or user identifier according to system logic. Address “any” signifies ::/0 and 0.0.0.0/0
simultaneously. (To use geolocation effectively you should download and install updates to the geolocation database at least
monthly—refer to SOL11176. IP addresses are linked to user identifiers in the BIG-IP’s IF-MAP server.)
A symbolic address is a fully-qualified domain name (FQDN) that a DNS resolver translates to one or more IP addresses which are
then used to match packet addresses. Symbolic addresses let you tie security rules to named services or logical hosts rather than
to ephemeral IP addresses (as commonly seen in cloud computing), thereby minimizing rule churn. The BIG-IP system refreshes
FQDN-to-IP-address translations at intervals controlled by an adjustable timer. Note that the IP addresses discovered for each FQDN
will be used during the whole refresh interval even if their DNS time-to-live (TTL) values are shorter. To avoid having rules with symbolic
(FQDN) addresses match packets against stale IP addresses, coordinate your network firewall FQDN Resolver Refresh Interval with
the TTL values in the DNS address resource records (A and AAAA RR's) for your FQDN's. Symbolic addresses are supported by
TMOS v12.0 and later. To view the current IP-address translations for an FQDN, issue a command of the following form (replace xxxx
with an FQDN like "example.net" or with "all"):
tmsh show /security firewall fqdn-info fqdn xxxx
F5 Deployment Guide
54
Data Center Firewall
Symbolic addresses are convenient but using them expands the attack surface of your firewall to include your DNS service.
Information about firewall rules will leak to adversaries who snoop on DNS traffic. An adversary who can alter or spoof DNS
entries can effectively modify (some of) your firewall rules. When using symbolic addresses you should: (1) secure your own DNS
nameservers against unauthorized access (including via dynamic DNS updates); (2) configure BIG-IP AFM for more secure DNS
resolution; (3) provide a secure network path (such as a management-access-only subnet and VLAN) between your BIG-IP devices
and any local nameservers they access; and (4) implement DNSSEC to secure the RR's for your symbolic addresses. By default,
AFM uses an insecure BIG-IP Network DNS Resolver which sends queries directly to the Internet. Consult Appendix C: Securing AFM
Domain Name Resolution on page 66 for instructions on configuring more secure DNS resolution for BIG-IP AFM.
An address with a route-domain identifier (%ID) appended will match packet addresses in the indicated route domain; addresses
without route-domain ID’s will match packets from any Route Domain. You can use rules with addresses lacking route-domain
identifiers to filter just packets received from a particular route domain by putting those rules in a policy applied to the route-domain
object. A policy on a route-domain object only ever sees packets from that route domain.
When writing rules be mindful that address “any” matches both IPv6 and IPv4 addresses even if all the other rules in some policy or
rule-list refer to, say, IPv4 only.
Port lists collect port numbers for use in one or more rules. Port lists may be nested, which is handy for putting names to port
numbers.
A rule which specifies a VLAN matches only packets that enter the BIG-IP from that VLAN. All AFM firewall rules are “ingress” rules.
If you put a rule in disabled state it will not match any traffic. You may enable/disable rules on a time schedule.
Use action Reject only in rules which discard packets arriving from your intranet (including business-partner networks via secure
VPN links) with intranet source addresses. Always “Drop” unwanted packets from the Internet (or from bogus source addresses!).
Rejecting unwanted packets helps you find and correct configuration problems in your intranet. However, you cannot fix configuration
problems outside your network. If you Reject packets from the Internet you cause two problems. First, you let bad actors launder
attacks through your facilities just by forging packet source addresses—when you Reject such packets, you actually send useless
RST or similar packets to the real owners of spoofed source addresses (annoying them and damaging your own reputation). Second,
you enable DoS attacks to waste outgoing bandwidth as well as incoming!
When a firewall rule matches a packet it may invoke a specified iRule. That iRule may override the firewall-rule’s action (thus
determining the policy result) and/or do something like log a message or tinker with subsequent processing. An iRule may inspect a
packet deeply. For example, an iRule could check some higher-level protocol data inside a UDP packet to decide whether to send it
to an alternate listener (virtual server) after accepting it. Since iRules are handy for logging packet characteristics (e.g., length) other
than normal matching criteria, you may configure a rule to invoke an iRule only for a sample (say, 1%) of packets matched.
You may wish to disable logging on rules which match a lot of benign packets. You probably should log (and react to) indications of
configuration errors or APT activity in your intranet—including packets with bogus source addresses reaching the firewall from the
intranet and attempted connections to Internet addresses with bad reputations.
Defining and Attaching Network Firewall Policies
You may create and edit a firewall policy before (or while) enforcing or staging it on an object. You may attach a given policy to more
than one object. (The management port is exceptional.) Changing the policy on an object will not interrupt any flows established
under a previous (named or implicit) policy, even if those flows could not be initiated under the new policy.
To create a new firewall policy using the TMOS 11.6 management GUI, navigate to Security > Network Firewall > Policies and click the
Create button. To edit an existing policy, simply click its name in the list.
After you create a policy you may enforce or stage it on a route domain, virtual server, or self IP object from the Security policies tab
for that object (e.g., Network / Route Domains / Route Domain 0 / Security).
The BIG-IP management GUI support for global-context firewall policy is a bit confusing in TMOS 11.6. To enforce or stage a policy
on the global object, navigate to Security / Network Firewall / Active Rules, then click the word “Global” at the left end of the very first
line in the list (ignore everything else on that line). The Security / Network Firewall / Active Rules / Global Firewall Rules page will open.
To enforce a policy go to the Network Firewall block, set “Enforcement” to “Enabled”, then set “Policy” on the same line to the policy’s
name (do not choose “Create”). To stage a policy, set “Staging” to “Enabled” and “Policy” on the same line to the policy name (do not
choose “Create”). Then click the Update button. Beware of the rule list at the bottom of the page. Use it to review rules (including
F5 Deployment Guide
55
Data Center Firewall
their status and hit counts) but not to change rules. Do not click the “Add” button on this page. Note: the “Policy Type” setting at the
top of the list changes which policy (enforced/staged) you see.
The single permanent network-firewall policy for the BIG-IP management port is found at System / Platform / Security. For important
advice about changing it, review Protecting BIG-IP Management Ports with AFM elsewhere in this document.
Optimizing Network Firewall Policies
For best system performance you may optimize your network firewall configuration in several ways: (1) when possible replace
duplicated data in rules and policies with references to address lists, port lists, and rule lists; (2) specify port ranges and address
ranges instead of listing adjacent ports or addresses (that is, use I–N rather than I,J,K,L,M,N); (3) when possible summarize adjacent
subnets or addresses as CIDR blocks (e.g., 10.0.0.0/24 and 10.0.1.0/24 become 10.0.0.0/23); (4) when you have two (or more)
rules which specify the same source and different destinations (or vice-versa), and the different destinations (or sources) could be
summarized using address and/or port blocks or ranges, consolidate those rules; (5) check (in the Active Rules List) for and remove
or correct overlapping and conflicting rules; and (6) remove unused address/port/rule lists and firewall policies from the BIG-IP
configuration.
IP Intelligence
AFM IP Intelligence offers enhanced capabilities for dynamic address matching. Like network firewall rules that match packet
addresses against geolocation or user identifiers, IP Intelligence policies match packet source addresses against blacklist categories.
(In TMOS 11.6 matching packet destination addresses against IP Intelligence blacklist categories requires an iRule like the example
afm-no-evil-dst in this document.) Categories represent dynamic lists of (subnet and host) addresses. IP Intelligence has eleven
blacklist categories built in (identifying botnets, DoS sources, address-concealing proxies, and so forth). Not every class of
untrustworthy IP addresses is represented by a built in category. You may add more local blacklist categories. For example, you
might add a local category to blacklist unallocated public addresses (“bogons”).
To build an IP Intelligence policy you select blacklist categories of concern, specifying for each the action—Drop, Accept, or “Use
Policy Default”—to take when a packet matches that category. It is best to choose “Use Policy Default”, which takes whatever
the policy’s “Default Action” (Drop or Accept) happens to be at the time the packet matches the category. The order of blacklist
categories in a policy does not matter. When a packet matches several categories it will be discarded if the action for any one of them
is Drop (i.e., Drop trumps Accept). (The main use of action Accept is to log the receipt of packets which match some category.)
You may simulate "staging" an IP Intelligence policy as follows: Set the policy’s Default Action to Accept and enable the two Default
Log Actions. Set the Action for each blacklist category to "Use Policy Default. Finally, attach the policy to a traffic-handling object to
log which packets the policy matches without actually blocking any traffic. When you are satisfied with the policy’s criteria change its
Default Action to Drop.
Setting the action for a blacklist category to Accept is not the same as defining a whitelist. There is only one whitelist per policy and
it does not use categories. An IP Intelligence policy will Accept any packet that matches a whitelist address, even if it also matches a
blacklist category (and regardless of any blacklist category action). You use blacklist categories to recognize and exclude connection
attempts from malicious actors or compromised networks (and sometimes just nuisances, like competitors trying to probe your
online catalog). You use whitelists to ensure that connections from trusted correspondents are welcomed even when their addresses
inadvertently match a blacklist category (probably because some other customer of their ISP is misbehaving).
There are two ways to add local blacklist categories but only one way to delete obsolete categories. You may add and delete
categories by editing the BIG-IP configuration (using TMSH or the management GUI at Security / Network Firewall / IP Intelligence /
Blacklist Categories). AFM automatically adds new local blacklist categories specified in address-list data (details below) but never
deletes any categories.
F5 supplies address data for the built-in blacklist categories on a near-realtime basis by subscription. The addresses are updated
frequently in response to activity on the Internet. For instance, when someone on a regional ISP network starts firing DoS attacks,
that will quickly be noticed and the afflicted network will be added to the relevant IP Intelligence category. To populate local blacklist
categories or policy whitelists with addresses you must configure one or more Feed Lists. Each Feed List periodically fetches
address data from HTTP(S) or FTP servers you specify.
IP Intelligence policies attached to global, route-domain, and virtual-server objects are enforced in that order. There is no default IP
Intelligence policy and self IP’s can’t have separate policies (though the global policy protects them). You cannot override a blacklist
Drop action in one policy with an Accept action or a whitelist match in a subsequent policy. Neither category nor whitelist addresses
F5 Deployment Guide
56
Data Center Firewall
may include Route Domain identifiers. (To match addresses in some route domain, attach an IP Intelligence policy to the routedomain object.)
Generally speaking you should apply a global “master” IP Intelligence policy to discard traffic matching most blacklist categories.
Virtual-server policies are mainly useful when you wish to exclude public proxy users or local nuisances from certain services only. In
that case, omit the “proxy” or “nuisance” blacklist category from the master policy but add it to a secondary policy, then attach the
secondary policy to specific virtual servers.
If you subscribe to the F5 IP Intelligence service you should Drop incoming packets that match any malice-associated blacklist
category. You might ignore the category “Cloud Provider Networks” since many benign connections originate from such addresses.
The “Proxy” category also deserves special consideration. Some bad actors use proxies to evade well-deserved shunning of their real
addresses, but legitimate actors may also use proxies (for instance, to dodge political controls on Internet usage in some countries).
It is wise to whitelist the public addresses of business partners with whom you maintain machine-to-machine links (EDI partners,
for example). Don’t forget to include the Feed List(s) for your whitelist(s) in any IP Intelligence policy which guards a resource any
business partner needs to access. The way IP Intelligence works, a whitelist match in a global policy will not avert or bypass a
blacklist-category match in a virtual-server policy (i.e., there is no IP Intelligence action equivalent to the network firewall action “Accept
Decisively”). Always whitelist remote VPN endpoint addresses. Also whitelist any public IP addresses you control (your IPv6 PI prefix,
for example).
Once you attach a global IP Intelligence policy you may test how it judges some address by issuing a command of the form:
tmsh show /security ip-intelligence info address xxxx
where xxxx is something like 203.0.113.9 or 2001:db8:cc0f:2152::9. To test a route-domain or virtual-server IP Intelligence policy
append “virtual server VVV” or “route domain /Common/DDD” to the command.
How To Create An IP Intelligence Policy
In TMOS 11.6 the procedure to create an IP Intelligence policy with a whitelist is not perfectly intuitive. Here is the recommended
approach:
First identify the web or FTP server from which a Feed List will retrieve the policy’s whitelist addresses. Place a whitelist address-list
on the server and note the URL needed to fetch it. (The address-list format is explained below. If you lack a suitable server, Appendix
B: Storing IP Intelligence Address Lists in BIG-IP Data Groups on page 62 shows how to supply addresses to Feed Lists from
BIG-IP data groups.) Next, create an IP Intelligence Feed List at Security / Network Firewall / IP Intelligence / Feed Lists and add a
Feed URL with “List Type” set to “White List” using the URL of your whitelist address list. Set “Poll Interval” to 600 seconds. (To
merge whitelist addresses from several sources, add more Feed URL’s with List Type set to White List.)
If the policy will include local blacklist categories (or if you plan to add addresses from non-F5 sources to built in categories), add one
or more Feed URL’s for blacklist addresses to the Feed List. Set each blacklist Feed URL’s “List Type” to “Black List” and select the
“Blacklist Category” which addresses from that Feed URL should populate by default (read down for details). You may wish to create
extra Feed Lists to segregate whitelist and blacklist Feed URL’s. Set each Feed URL’s Poll Interval to the largest value consonant with
the source data’s update schedule.
Next, create the IP Intelligence policy at Security / Network Firewall / IP Intelligence / Policies. Select the Feed List(s) which will
retrieve whitelist addresses and populate or augment the blacklist categories you will use in the policy. If you want to use a whitelist
in an IP Intelligence policy you must select at least one Feed List. If you would like to “stage” the policy, set Default Action to Accept
(otherwise leave it set to Drop). For each blacklist category you wish to filter in this policy: select the category name; leave the percategory Action and Log choices as “Use Policy Default”; click Add. (Note that to use a local category in a policy you must also select
Feed List(s) to populate it. Otherwise the category will be empty and no packets will match it!) Finally, click Finished.
(Every address whitelisted by some Feed URL Z will be added to the whitelist of any policy which selects a Feed List that includes Z.)
IP Intelligence Address Data Format
AFM expects Feed URL’s to retrieve address lists in Comma-Separated Value (CSV) format. Each record (line) must have an IP
address in the first field, then optionally: a prefix/subnet mask-length in the second field; a whitelist indicator in the third field; and a
blacklist category identifier in the fourth field. Any or all fields after the first may be empty. An empty mask-length means the address
identifies a single host.
F5 Deployment Guide
57
Data Center Firewall
The third and fourth fields (whitelist indicator and blacklist category name) are interpreted as follows: first, if the third field is empty a
value of “wl” or “bl” is presumed: “wl” if the Feed URL’s “List Type” is “White List” and “bl” otherwise.
Then, if the third field’s value is “wl” or “whitelist” the address is whitelisted and the fourth field is ignored. If the third field’s value
is “bl” or “blacklist” the address is added to the blacklist category named in the fourth field, or if that field is empty, to the Feed
URL’s Blacklist Category (any other value in the third field makes AFM ignore the whole line). When the third field’s value is “bl” and
the blacklist category named in the fourth field does not exist, AFM creates that blacklist category automatically up to a limit of 62
categories.
Feed URL’s can fetch mixed lists of whitelist and blacklist-category addresses but it is better to separate the two types of addresses.
When you expect a Feed URL to retrieve mainly whitelist (or blacklist) addresses, set its List Type to White List (Black List). That will
help people understand the BIG-IP configuration even if the addresses actually come with whitelist indicators. However, when you
expect a Feed URL to fetch mixed addresses set its List Type to Black List. That won’t affect the way address records with whitelist
indicators are processed but will ensure that addresses lacking whitelist indicators are not whitelisted by default. It is much more
dangerous to whitelist an address by mistake than to blacklist it.
F5 Deployment Guide
58
Data Center Firewall
Denial-of-Service (and DNS) Protection
This section provides additional information about DoS and DNS protection.
The Denial-of-Service Landscape
AFM DoS Protection mitigates two main types of DoS (DDoS) attacks: processing-logic attacks and resource-exhaustion attacks (the
types do overlap).
Processing-logic attacks involve sending packets or messages (attacks are possible at every protocol layer) which are malformed in
some way (sometimes in a way which only matters to a few possible recipients). Ideally any machine receiving a malformed message
would simply reject it but in practice many recipients react badly—some even crash. In the worst case recipients react by sending
replies which the attacker can direct to other victims by address spoofing.
Resource-exhaustion attacks involve sending well-formed packets or messages which ask the recipient to expend some resources.
The more packets the attacker sends the more resources are wasted, leaving less for legitimate correspondents. These attacks
may be subtle or crude. The well-known SYN-flood attack is fairly subtle. Only a few thousand packets are required to disrupt a
vulnerable device because each provokes a big waste of memory. So-called volumetric attacks are cruder: the attacker just sends (or
uses an amplification attack to make lesser victims send) very large numbers of packets. Indeed, the biggest DoS attacks commonly
aim to exhaust the victim’s ISP link capacity. Under such a DoS attack it doesn’t matter how efficiently the victim’s firewall recognizes
and discards malicious packets—they crowd-out legitimate packets upstream.
AFM DoS Protection inspects packets and messages. It discards those which are malformed then recognizes and rate-limits those
which are likely subtle resource-exhaustion attacks. DoS Protection also rate-limits volumetric attacks. LTM and AFM also implement
specialized defenses to some attacks of the latter kind—for example, using SYN cookies to mitigate SYN floods.
The only defense to a really big DoS attack is cloud scrubbing. Cloud scrubbing means diverting traffic addressed to the victim to
a facility with enormous ISP bandwidth and firewall capacity which “scrubs” malicious packets from the datastream then sends the
residuum of legitimate packets on to the victim. F5 offers a cloud-scrubbing service called F5 Silverline DDoS Protection (Silverline). If
your organization might be subjected to a volumetric DoS attack you should consider a Silverline on-demand subscription. You may
configure your BIG-IP data center firewall to recognize DoS attacks and activate Silverline cloud scrubbing when needed. That way
you minimize network latency in the normal case but enjoy massive defensive capacity when required.
The majority of DoS attacks are not so big. In fact, almost all organizations log a constant stream of petty DoS attacks as malicious
actors probe target networks more or less randomly. Bad actors are always looking for minor victims they can exploit to mount
amplification attacks against other targets. Note that AFM IP Intelligence and network firewall provide critical defenses against DoS
attacks: IP Intelligence empowers you to discard packets from malicious senders without wasting system resources trying to analyze
them. The network firewall discards packets with bogus addresses—nearly all of them are attacks.
Understanding AFM DoS Protection
The first element of AFM DoS Protection is called (somewhat obscurely) Device Configuration. Device Configuration applies a global
DoS-protection policy to all traffic through the BIG-IP system except management traffic. It recognizes malformed (e.g., Christmastree) and otherwise suspicious packets that match patterns called “DoS vectors.” It also analyzes the arrival rates of various packet
types to distinguish likely volumetric DoS attacks from normal traffic. Log messages from DoS Device Configuration alert you to DoS
attacks and help you tune policy criteria to mitigate them.
DoS Device Configuration simply discards most malformed packets (at wire speed with F5 BIG-IP hardware acceleration ). It accepts
merely suspicious packets for further processing unless it sees too many of them in a short time, in which case it rate-limits similar
packets until the presumed attack subsides.
How many is too many? The answer depends on the characteristics of your network and applications. You should tune the policy
(Security > DoS Protection > Device Configuration) of quantity and rate-of-change thresholds for detection of various DoS vectors.
The default criteria represent F5’s best judgment for a “starter policy,” which is essential because the DoS Device Configuration
policy is always enforced. (Note that the starter policy imposes no rate limit at all on some packet types (UDP, for example). You are
encouraged to make adjustments.)
Some applications are more sensitive than others to DoS attacks of various sorts. You may wish to protect a sensitive application
with a “tight” policy, but if you tighten the global DoS Device Configuration policy too much it may interfere with traffic to other, less
sensitive or simply busier applications. To avoid that problem you can utilize the second element of AFM DoS Protection, the DoS
Protection Profile.
F5 Deployment Guide
59
Data Center Firewall
Each DoS Protection Profile defines a DoS-protection policy for some application or service. DoS Protection Profiles include general
criteria and special criteria for DNS and SIP traffic. To enforce a DoS Protection Profile you must attach it to a virtual server. The
quantity and rate-of-change thresholds for various vectors are then applied to detect and mitigate DoS attacks in traffic through that
virtual server specifically. (DoS Protection Profiles don’t see obviously-malformed packets because those are discarded early by DoS
Device Configuration or by LTM, which implements many network protocols very exactly.)
When you protect specific applications and services with Dos Protection Profiles you may tune DoS Device Configuration policy to the
overall contours of traffic through your network.
This Deployment Guide cannot offer a better example of DoS Protection policy than the default DoS Device Configuration supplied
with BIG-IP AFM. Review the AFM DoS Protection documentation for additional insight into this important BIG-IP AFM feature.
Protecting DNS
Most attacks on DNS can be classified as DoS attacks (though intrusive attacks are possible; see CVE-2001-0010 for an example).
You may use AFM to mitigate many DoS threats by discarding packets with bogus source addresses (network firewall, IP Intelligence),
rate-limiting malformed packets, bad queries, and query floods (AFM DoS Protection), and filtering out query types your servers don’t
support (DNS Protocol Security). However, many adversaries will try to overwhelm your DNS with a huge volume of superficially-valid
queries. To distinguish malicious queries from legitimate ones you must process them, incurring the exact cost you wish to avoid.
You cannot safely rate-limit queries because that facilitates cache-poisoning attacks against legitimate clients. The most effective
mitigation measure is simply to deploy an ultra-high-performance DNS intermediate server: F5 BIG-IP DNS Express.
DNS Express is part of the F5 BIG-IP Global Traffic Management and DNS Services module (GTM). GTM is a remarkably useful
product which you may configure with AFM and LTM to achieve comprehensive DNS security. However, the details are beyond the
scope of this Deployment Guide.
Due to the variety of DNS architectures implemented by F5 customers this document cannot provide definitive guidance on deploying
AFM with DNS. Basically, if you expose nameservers to the public Internet, you should admit queries via LTM virtual servers with DNS
Profiles, let your AFM global network-firewall and IP Intelligence policies scrub off packets with bad source addresses, tune an AFM
DoS Protection Profile to your DNS traffic and attach it to your DNS virtual servers, and if appropriate, configure and attach a DNS
Protocol Security Profile as well.
Troubleshooting AFM
The chief symptom of a firewall configuration error is lack of connectivity to some resource. That is also the chief symptom of many
another infrastructure or resource (server) problem. General network troubleshooting is beyond the scope of this document. The
advice which follows assumes you have good reason to suspect a firewall problem.
First check if BIG-IP service monitors can talk to the resource in question. Since they initiate flows from the BIG-IP itself they are
never blocked by firewall policy. You may also try to reach the resource from the BIG-IP command line. If a monitor cannot talk to a
resource, fix that before digging into firewall configuration.
If you suspect a non-firewall problem but your colleagues are skeptical, you might hazard a pinhole rule for service proof. Note the
address of a test client plus the address and port of the target resource. Find the network firewall policy enforced on the global object.
Prepend a rule of the form “src clientIP:any dst tgtIP:tgtPort vlan any action Accept_Decisively log Enabled” to that policy (temporarily,
of course). Also whitelist the test client’s address in any IP Intelligence policy along the path to the resource. Test connectivity. If the
test client still cannot talk to the resource, that points toward a non-firewall issue. (If your pinhole rule alleviates the problem, brag that
you have narrowed things down a lot, then extend your analysis.)
Examine the logs. Look for logs matching both the client and resource addresses. In many cases log messages will indicate which
firewall policies and rules (if any) are interfering with access.
If AFM is running in Firewall mode, check that the traffic-handling objects in the path to the resource have suitably-permissive policies
enforced on them. In some situations it pays to move a policy you suspect is causing trouble to staging and temporarily enforce a
policy like /Common/accept-all while you analyze the logs.
If the complaint involves intermittent connectivity, slowness, or excessive retransmissions when accessing a popular resource, check
whether a DoS Protection policy is rate-limiting traffic to the resource. If not, consider the possibilities of route-flapping or some-such
elsewhere in the network. Also consider load-balancing problems, like a pool member which responds to the service monitor but not
to actual application clients.
F5 Deployment Guide
60
Data Center Firewall
You may test the network-firewall policy hierarchy from the TMSH command line. Note the client’s source address (and port, if
possible—many connections are initiated from random ports above 1023 so you may have to choose one yourself; 9999 is easy to
type). Note the address and port of the target resource (as seen by the client; i.e., a virtual-server address and port in many cases).
Note as well the protocol (typically TCP or UDP) and the VLAN (like /Common/internal) from which client packets reach the BIG-IP.
Then invoke a command of this form (line breaks inserted here for readability—don’t type them into the BIG-IP):
tmsh show /security firewall matching rule
source-addr Q.R.S.T source-port NNN
dest-addr W.X.Y.Z dest-port MMM
protocol PPP vlan LLL
You will recall the existence of a similar command for checking IP Intelligence policy results:
tmsh show /security ip-intelligence info address W.X.Y.Z
where only the client’s address matters. To test a route-domain or virtual-server IP Intelligence policy append “virtual server VVV” or
“route domain /Common/DDD” to that command.
To review the current IP address translations for a symbolic (FQDN) address in a firewall rule enter a command of this form (replace
xxxx with the FQDN like "example.net" or with "all"):
tmsh show /security firewall fqdn-info fqdn xxxx
If all the symbolic address translations are missing, check and ensure that you configured a management default route or specific
route(s) to your DNS server(s).
F5 Deployment Guide
61
Data Center Firewall
Appendix B: Storing IP Intelligence Address Lists in BIG-IP Data Groups
The only way to populate addresses into IP Intelligence whitelists and blacklist categories in TMOS 11.6 is through Feed Lists.
However, it is not always convenient to set up a web or FTP server for IP Intelligence Feed URLs to poll. A workaround is possible.
The combination of virtual server and iRule shown below lets you make address lists available to Feed URLs from BIG-IP data groups.
You may use the management GUI or TMSH to edit records in “internal” data groups (Local Traffic > iRules > Data Group List), or you
can edit address lists elsewhere and upload them to “external” datagroups (System > File Management > Data Group File List).
BIG-IP data groups of type Address hold records with two fields, an IP address in CIDR notation and a string value. As detailed
above in IP Intelligence Data Format, Feed URLs consume records with two to four fields. The data group address field fills the
first two Feed URL fields exactly. The data group string value stores data for the remaining two fields plus tag information for the
convenience of administrators.
Compose each address record like this:
address/mask-length := [+][tag][=category]
where an optional leading ‘+’ indicates the address should be whitelisted, an optional tag names the address to help administrators
maintain the list, and an optional category (preceded by ‘=’) says to which blacklist category the address belongs. For example:
10.0.0.0/8
2001:db8:cc0f::/48
2001:db8:ffff::/48
:= +intranet
:= Mythical Medical Ctr=shun
:= =bogons
Since all the data in the string value is optional, string values may be empty. If the first character of the string value is ‘+’ the third field
of the Feed URL record will be “wl”. If present, the category will fill the fourth field of the Feed URL record. The tag will not appear in
the Feed URL record.
To serve address lists from data groups to multiple BIG-IP’s (and/or other clients) put your datagroup-to-address-list (D2AL) virtual
server in traffic-group-1 and give it an intranet (locally-routeable) address. If all of your BIG-IP’s belong to one Device Group you may
create a D2AL virtual server on each BIG-IP in traffic-group none and rely on config-sync to replicate your address-list data groups.
This example shows the second approach (one D2AL virtual server on each BIG-IP at a non-routeable address).
Create the iRule on the following page to format Data Group records for Feed List Feed URLs.
F5 Deployment Guide
62
Data Center Firewall
iRules (Navigate to Local Traffic > iRules)
Name
Give the unique name. We use afm datagroup feedurl
Definition
Copy and paste the following iRule, omitting the line numbers:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
F5 Deployment Guide
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
#
In TMOS 11.6, IP Intelligence blacklist and whitelist addrs
may be updated by IPI Feed Lists but not from TMSH or TMGUI.
It is easy, though, to edit an internal Address datagroup...
Create at least one "Address" datagroup (internal or external;
edit internals from Local Traffic/iRules/Data Group List) to
list IP's and (optional) values. E.g., /Common/graylist.
When given, values are "[+][TAG][=CAT]" where optional TAG is
address owner and optional CAT is IPI category. Optional '+'
indicates address should be whitelisted. Examples:
10.0.0.0/8
:= "+intranet"
203.0.113.0/24 := "example.com=phishing"
Create a virtual server with HTTP Profile on an IP reachable
by intended list users. Attach this iRule. No pool. Set a
Feed List source URL like "http://198.18.0.5/Common/graylist"
with datagroup’s full /Partition/name as URL path. This iRule
reports datagroup records as CSV lines: "addr,masklen,W,CAT"
where W or CAT may be empty. For example:
10.0.0.0,8,wl, | 203.0.113.0,24,,phishing
URL queries can select records by TAG and/or CAT:
http://198.18.0.5/Common/graylist?t=example.com
http://198.18.0.5/Common/graylist?c=phishing
when HTTP_REQUEST {
set dg [HTTP::path]
if {([HTTP::method] ne "GET") || ![class exists $dg] ||
([class type $dg] ne "IP") || ![class size $dg]} {
log local0.err "[IP::client_addr] bad query to ${dg}"
HTTP::respond 404 ; return
}
set buf [set c [set t ""]]
foreach {q} [split [HTTP::query] "&"] {
foreach {n v} [split $q "="] {
if {($n eq "t") || ($n eq "c")} { set $n $v }
}
}
foreach {item} [class get $dg] {
foreach {ip val} $item {
set w ""
if {[string range $val 0 0] eq "+"} {
set w "wl" ; set val [string range $val 1 end]
}
foreach {addr len} [split $ip "/"] {
foreach {tag cat} [split "${val}=" "="] { break }
if {(($t ne "") && ($tag ne $t)) ||
(($c ne "") && ($cat ne $c))} { break }
append buf "${addr},${len},${w},${cat}\n"
}
}
}
set ct "text/plain; charset=UTF-8"
HTTP::respond 200 content $buf Content-Type $ct
}
63
Data Center Firewall
The next task is to add a virtual server that Feed URLs can poll.
Unless you choose to assign a routeable intranet address to this virtual server, create it separately on each BIG-IP device (big-s1 and
big-s2 in our example) and move its virtual address to traffic-group none.
Virtual Servers (Navigate to Local Traffic > Virtual Servers)
Name
datagroup-feedurl-vs
Destination Address
198.18.0.5 (you could use an intranet address; see the note preceding this table)
Service Port
80 (HTTP)
HTTP Profile
http
VLAN and Tunnel Traffic
Select Enabled On, and then select only the Internal VLAN you created.
Source Address Translation
None
iRules
Enable the iRule you just created (afm-datagroup-feedurl in our example).
Default Pool
None
Modifying the Traffic Group
If the datagroup-feedurl-vs you just created has a non-routeable address, you must move its virtual address to traffic-group none. If
using a routeable intranet address leave it in traffic-group-1.
If necessary, click Local Traffic > Virtual Servers. On the menu bar, click Virtual Address List. Click the IP address of the virtual
server you just created (198.18.0.5 in our example). From the Traffic Group list, select None, and then click Update.
Creating blacklist categories and feed lists
The next task is to create blacklist categories and then Feed Lists to populate them. Use the following table for guidance.
Black List Categories (Navigate to Security > Network Firewall > IP Intelligence > Black List Categories)
Black List Category Name
shun
Description
Public IP addresses of nuisances or competitors we want to keep out of our websites.
You will keep local blacklist and whitelist addresses in data groups. Internal data groups (as shown here) can hold a few thousand
records, but are not convenient for more than fifty or so. Consider using external data groups instead. Just maintain your address
lists with a text editor and upload them to the BIG IP system for distribution.
iRule Data Group (Navigate to Local Traffic > iRules > Data Group List)
AFM Blacklist Data Group
Name
afm-blacklist
Type
Address
Records > Address
Address
2001:db8:cc0f::/48
Value
Mythical Medical Ctr=shun
AFM Whitelist Data Group
Name
afm-whitelist
Type
Address
Records
F5 Deployment Guide
Address
Value
192.0.2.0/24
Example.net (whitelist our own public addresses)
2001:db8:16d::/48
Example.net (whitelist our own public addresses)
170.167.0.0/16
Costco Wholesale
64
Data Center Firewall
Feed Lists (Navigate to Security > Network Firewall > IP Intelligence > Feed Lists)
Local blacklist Feed List
Name
local-blacklist-feed
Description
local blacklist feed
Feed List Properties: Feed URLs
Name
local-blacklist
URL
http://198.18.0.5/Common/afm-blacklist
List Type
Black List
Blacklist Category
shun
Poll Interval
600
Local whitelist Feed List
Important Note
This is the white list feed list referenced in Creating the Local whitelist feed on page 22. If you have already created that
Feed List, simply edit the existing list to add the URL referenced below.
Name
local-whitelist-feed
Description
Trusted addresses for our company and our business partners.
Feed List Properties: Feed URLs
Name
local-whitelist
URL
http://198.18.0.5/Common/afm-whitelist
List Type
White List
Blacklist Category
spam_sources (What you choose here does not matter; it can be anything)
Poll Interval
600
Updating the IP Intelligence Policy
The final task is to update the IP Intelligence policy you created in Creating an IP Intelligence policy on page 22 (in this example, the
one on the global object) to use the objects you just created.
IP Intelligence Policy (Navigate to Security > Network Firewall > IP Intelligence > Policies)
Important Note
This is the IP Intelligence policy referenced in Creating an IP Intelligence policy on page 22. If you have already created
that policy, simply edit the existing policy to add the local blacklist feed and shun category you just created.
Name
global-IPI-policy
Description
Prevents connections from bogus and malicious source addresses
IP Intelligence Properties
Feed Lists
Use the Add arrows (<<) to move the four Feed Lists you created (bogons feed,
DROP_EDROP feed, local-whitelist-feed, and local-blacklist-feed) to the Selected
box. Note that the local-blacklist-feed is the one you just created.
Default Action
Drop
Default Log Actions
Check the Log Blacklist Category Matches and Log Whitelist Overrides boxes.
Blacklist Matching Policy
From the Blacklist Category list, select each of the following categories and then
click Add (leave the Action and both log lists at the default (Use Policy Default)
bogons
botnets
denial_of_service
DROP_EDROP
illegal_websites
infected_sources
phishing
proxy
shun (Note this is the category you just created)
scanners
spam_sources
web_attacks
windows_exploits
F5 Deployment Guide
65
Data Center Firewall
Appendix C: Securing AFM Domain Name Resolution
In TMOS version 12.0 the BIG-IP AFM network firewall resolves symbolic (FQDN) address domain names to IP addresses using a
BIG-IP Network DNS Resolver. A TMOS 12.0 Network DNS Resolver sends DNS queries directly to Internet nameservers by default,
but you can make it query local nameservers by configuring a suitable Forward Zone. TMOS 12.0 Network DNS Resolvers do not
implement DNSSEC so they are unable to validate responses to their queries cryptographically.
To reduce the danger that an adversary will subvert your AFM network firewall rules by interfering with Internet DNS resolution, you
should use a Network DNS Resolver Forward Zone to direct AFM’s DNS queries to a BIG-IP DNS Cache Validating Resolver or to
local nameservers which use DNSSEC to cryptographically-validate DNS responses.
If your BIG-IP device has a suitable license, you may configure a BIG-IP DNS Cache Validating Resolver to supply cryptographicallyvalidated DNS responses to AFM’s Network DNS Resolver. This approach is the most secure since it excludes the local network from
your firewall attack surface (to the extent that symbolic addresses in AFM network firewall rules refer to cryptographically-secured
(DNSSEC) DNS resources).
Otherwise, if you have DNSSEC-enabled local nameservers and can provide secure network paths between them and your BIG-IP
device(s) you may obtain a pretty good level of security by directing Network DNS Resolver queries to those nameservers. Since
responses to those queries will not be cryptographically-secured in transit, your firewall attack surface will still include network paths to
your local nameservers.
We do not recommend allowing AFM’s Network DNS Resolver to send queries directly to the Internet.
Configuring a BIG-IP DNS Cache Validating Resolver
If your BIG-IP device has a suitable software license, follow these instructions to configure a DNS Cache Validating Resolver.
Otherwise, skip to Using Secure Local Nameservers on page 69.
On the day you configure the DNS Cache Validating Resolver you must obtain some fresh DNSSEC public keys. You will need the
long-term DNS root key-signing-key(s) (KSK’s), the long-term dlv.isc.org KSK(s), and optionally the long-term DLV KSK(s) for your local
secure DNS zone(s). Use the dig(1) command from the BIG-IP Advanced Shell command line (or from another workstation) to obtain
key-signing-key RR’s as follows (use the virtual console or an SSH client such as putty to access the BIG-IP Advanced Shell):
First obtain the long-term root KSK RR(s):
dig -t dnskey . | grep 'IN.DNSKEY.257'
You will see something like this example:
.
107052 IN
DNSKEY 257 3 8
AwEAAagAIKlVZrpC6Ia7gEzahOR+9W29euxhJhVVLOyQbSEW0O8gcCjF FVQUTf6v58fLjwBd0YI0EzrAcQqBGCzh/RStIoO8g0NfnfL2MTJRkxoX
bfDaUeVPQuYEhg37NZWAJQ9VnMVDxP/VHL496M/QZxkjf5/Efucp2gaD X6RS6CXpoY68LsvPVjR0ZSwzz1apAzvN9dlzEheX7ICJBBtuA6G3LQpz
W5hOA2hzCTMjJPJ8LbqF6dsV6DoBQzgul0sGIcGOYl7OyQdXfZ57relS Qageu+ipAdTTJ25AsRTAoub8ONGcLmqrAmRLKBP1dfwhYB4N7knNnulq
QxA+Uk1ihz0=
That is the long-term root key-signing-key RR. If you see two RR’s that means the root KSK will be replaced soon. Copy and save
the key RR(s) to use momentarily.
Note: the value shown in this document is just an example. Use the dig(1) command to obtain fresh key RR(s).
Next obtain the long-term dlv.isc.org key RR:
dig -t dnskey dlv.isc.org. | grep 'IN.DNSKEY.257'
You will see something like this example:
dlv.isc.org.
7200
IN
DNSKEY 257 3 5
BEAAAAPHMu/5onzrEE7z1egmhg/WPO0+juoZrW3euWEn4MxDCE1+lLy2 brhQv5rN32RKtMzX6Mj70jdzeND4XknW58dnJNPCxn8+jAGl2FZLK8t+
1uq4W+nnA3qO2+DL+k6BD4mewMLbIYFwe0PG73Te9fZ2kJb56dhgMde5 ymX4BI/oQ+cAK50/xvJv00Frf8kw6ucMTwFlgPe+jnGxPPEmHAte/URk
Y62ZfkLoBAADLHQ9IrS2tryAe7mbBZVcOwIeU/Rw/mRx/vwwMCTgNboM QKtUdvNXDrYJDSHZws3xiRXF1Rf+al9UmZfSav/4NWLKjHzpT59k/VSt
TDN0YUuWrBNh
That is the long-term dlv.isc.org KSK RR. If you see two RR’s that means the KSK will be replaced soon. Copy and save the key
RR(s) to use momentarily.
Note: the value shown in this document is just an example. Use the dig(1) command to obtain fresh key RR(s).
F5 Deployment Guide
66
Data Center Firewall
If your organization has any local DNSSEC-secured zones which use DLV KSK’s not registered with dlv.isc.org, obtain RR’s for those
keys from your DNS administrator.
Configuring the BIG-IP objects
Use the following guidance to configure the required objects on the BIG-IP system. Each table contains only the required settings
for each configuration object. Unless otherwise stated, settings not listed in the table can be configured as applicable for your
implementation. For specific instructions on configuring the objects, see the inline help or product documentation.
Creating a new Cache
Use the following table to create a new Cache with a type of Validating Resolver.
Cache (Navigate to DNS > Caches > Cache List > Create)
Name
Type a unique name, such as fw-secure-resolver.
Resolver Type
Validating Resolver
Modifying the Cache to add a Trust Anchor
Next, you open the Validating Resolver object you just created and use the tabs to edit key properties. The first task is to add each
long-term root key-signing-key RR to Trust Anchors:
Cache: Trust Anchor (Click the Cache you just created > Trust Anchors (on the Menu bar) > Add)
Trust Anchor
Paste one entire long-term root key-signing-key RR which you obtained earlier using dig(1).
If you have more than one long-term root key-signing-key RR, click the Add button and insert the additional key.
Modifying the Cache to add a DLV Anchor
Next, add each long-term dlv.isc.org KSK RR to DLV Anchors:
Cache: DLV Anchor (Click DLV Anchor (on the Menu bar) > Add)
DLV Anchor
Paste one entire long-term dlv.isc.org key-signing-key RR which you obtained earlier using dig(1).
If you have more than one long-term dlv.isc.org key-signing-key RR, click the Add button and insert the additional key.
Repeat to add any long-term DLV KSK RR’s for your local secure zones to DLV Anchors.
i
Important With TMOS 12.0 you must maintain Trust and DLV Anchors for DNS Cache Validating Resolvers manually. When
the key-signing-key for the DNS root, the dlv.isc.org zone, or any local secure zone expires you must remove it
and insert the replacement KSK RR.
If your BIG-IP system does not have a route to send DNS queries to the Internet
If your BIG-IP device does not have a route to send DNS queries directly to the Internet, you may configure your DNS Cache
Validating Resolver to send all queries to local forwarding nameservers. You must add a special Forward Zone as described in this
section.
o not use a Forward Zone to make your DNS Cache Validating Resolver send all DNS queries to local nameservers
 Note: D
unless necessary. Replace the example nameserver addresses here with values appropriate in your environment.
In the following table, we show an example with two forwarding nameservers: 192.168.193.5 and 192.168.193.7.
Cache: Forward Zones (Click Forward Zones (on the Menu bar) > Add)
[that is a single dot ‘.’ by itself]
Name
.
Nameservers
Address:
192.168.193.5
Service Port: 53
Click Add to include more Nameservers. In our example, we add 192.168.193.7
Additionally, if you have any local zones which should be resolved by queries to specific local servers, you may add corresponding
Forward Zones. For example, you might have a zone corp.example.net which is only resolved by a local nameservers
192.168.193.5 and 192.168.193.7.
F5 Deployment Guide
67
Data Center Firewall
he following is just an example. Do not add any local Forward Zones unless you need them. Replace the example
 Note: T
data here with values appropriate in your environment.
Cache: Forward Zones (Click Forward Zones (on the Menu bar) > Add)
Name
corp.example.net (our example)
Nameservers
Address:
192.168.193.5
Service Port: 53
Click Add to include more Nameservers. In our example, we add 192.168.193.7
Creating a DNS profile
The next task is to create a DNS profile. Use the following table for guidance on configuring the profile.
Cache: Forward Zones (Click Forward Zones (on the Menu bar) > Add)
Name
Type a unique name, such as fw-secure-dns-pfl
Parent Profile
dns
DNS Cache
Enabled
DNS Cache Name
Name of the cache you created (fw-secure-resolver in our example)
Unhandled Query Actions
Reject
Use BIND Server on BIG-IP
Disabled
Process Recursion Desired
Enabled
Creating the virtual servers
The next task is to create the virtual servers on the BIG-IP system. To avoid IP address conflicts you may use a special "class E" IP
address valid only inside the BIG-IP system for these virtual servers. We suggest the one shown in the example.
Virtual Servers (Main tab > Local Traffic > Virtual Servers)
UDP virtual server
Name
Type a unique name, such as fw-secure-dns-udp-vs.
Destination Address
Type the IP address for this virtual server, such as 245.0.0.53/0.
Service Port
53
Protocol
UDP
DNS Profile
Name of the DNS profile you created (fw-secure-dns-pfl in our example)
VLANs and Tunnels
All VLANs and Tunnels
Source Address Translation
None
TCP virtual server
Name
Type a unique name, such as fw-secure-dns-tcp-vs.
Destination Address
Type the IP address for this virtual server, such as 245.0.0.53/0.
Service Port
53
Protocol
TCP
DNS Profile
Name of the DNS profile you created (fw-secure-dns-pfl in our example)
VLANs and Tunnels
All VLANs and Tunnels
Source Address Translation
None
F5 Deployment Guide
68
Data Center Firewall
Adding a network firewall policy to each virtual server
The next task is to attach an AFM network firewall policy to the virtual servers you just created.
Virtual Server (Navigate to Local Traffic > Virtual Servers > your-udp-virtual-server > Security (Menu bar) > Policies )
Navigation help
Click Local Traffic > Virtual Server. Click the UDP virtual server you created. On the Menu bar, click Security >
Policies.
Network Firewall: Enforcement
Select Enabled. From the Policy list that appears, select the Accept All policy you created in Creating the Network
Firewall Policies on page 18 (accept-all in our example).
Network Firewall: Staging
Select Disabled. .
IP Intelligence
Leave IP Intelligence set to Disabled.
DoS Protection Profile
Leave DoS Protection Profile set to Disabled.
Log Profile
None
Repeat this table for the TCP virtual server you created
When you configure the Network DNS Resolver Forward Zone in the following section, the IP address for your Forward Zone
nameserver will be the address you used for the DNS Cache Validating Resolver virtual servers, such as 245.0.0.53 in our example.
Using Secure Local Nameservers
If your BIG-IP device’s license does not include DNS Cache Validating Resolver capability, you may use secure local nameservers
to improve the security of AFM symbolic-address resolution. Ensure that your local nameservers are configured to handle recursive
queries from clients and to validate results using DNSSEC. Ensure the network paths between your BIG-IP AFM devices and your
nameservers are secure.
Configuring AFM to Resolve Symbolic (FQDN) Addresses in Network Firewall Rules
The next task is to create a Network DNS Resolver. In the following table, the small Cache Size value is appropriate because this
object will use external caches.
DNS Resolver (Navigate to Network > DNS Resolvers > Create)
Name
Type a unique name, such as fw-net-resolver.
Route Domain Name
0
Cache Size
131072
Use IPv6
Uncheck this option unless both of the following are true: you are using local secure nameservers (not a DNS Cache
Validating Resolver), and at least one of your local secure nameservers has an IPv6 address.
Adding a Forward Zone to the DNS resolver
The next task is to open the resolver you just created and add a single Forward Zone. The nameserver IP address(es) for that Forward
Zone will either be the one you assigned to the virtual servers for any DNS Cache Validating Resolver you created (e.g., 245.0.0.53)
or the IP addresses of your secure local nameservers (replace the example addresses 192.168.193.5/7 with the IP addresses of your
nameservers).
Forward Zone (Navigate to Network > DNS Resolvers > name of the DNS resolver you just created > Forward Zone (on the Menu bar) > Add
[that is a single dot ‘.’ by itself]
Name
.
Nameservers
When using a DNS Cache validating Resolver (actual IP)
Address:
245.0.0.53
Service Port: 53
When using secure local nameservers (example IP’s)
Address:
Service Port:
Address:
Service Port:
F5 Deployment Guide
192.168.193.5
53
192.168.193.7
53
69
Data Center Firewall
Next, you configure AFM to use your Network DNS Resolver (remember to click Update after making changes):
Network Firewall (Navigate to Security > Options > Network Firewall)
Global Context
Select the DNS Resolver you just created (fw-net-resolver in our example).
Refresh Interval
20 minutes.
If you use more than 200 different symbolic addresses in network firewall rules you may wish to lengthen the FQDN Resolver Refresh
Interval. Longer refresh intervals reduce BIG-IP CPU utilization but increase the chance that IP addresses in network firewall rules will
be stale. Consider the TTL values of your symbolic addresses when you adjust the refresh interval.
DNS Cache Validating Resolver Statistics
When you use a DNS Cache Validating Resolver you can check its performance. Navigate to DNS / Caches / Cache List / [name
of your Validating Resolver, such as fw-secure-resolver] then select the Statistics tab. Click the View link in the Details column
of the relevant row to review detailed information about queries, cache performance, resolver traffic, and DNSSEC coverage.
F5 Deployment Guide
70
Data Center Firewall
Glossary of Terms
AAA Gateway A device which enforces security policy very high in the OSI stack by authenticating users, authorizing
their connections, and managing their access by proxying or filtering traffic, possibly through dynamic
configuration of other infrastructure devices (from Authentication, Authorization, and Access Management )
ADC Mode
A BIG-IP AFM operating mode in which the implicit policy for every listener (virtual server or self IP)
accepts whatever traffic reaches it (from “Application Delivery Controller,” BIG-IP’s fundamental job)
ALG
Application Layer Gateway—a traffic-handling object which supports a complex network protocol.
A BIG-IP ALG for some protocol may create ephemeral listeners for secondary flows used in that
protocol and may discard packets which don't conform to that protocol.
APT Malware resident on a computer such as a virus, rootkit, or bot (from “Advanced Persistent Threat”)
AUP
Acceptable Use Policy—an organization’s rules for use of computing and Internet resources
Deep Packet Inspection
Examining packet payload, not just header and L2 context. Very sophisticated devices (e.g., BIG-IP)
can examine payloads split across multiple packets in a flow
Denial-of-Service
Protection
Component of BIG-IP AFM which recognizes and excludes network DoS attacks
DDoS
Distributed DoS
DLP
Data Loss Prevention—tool which scans network traffic looking for confidential data in payloads. May
log or interrupt data transfers which lack administrative approval
DMZ A small trust environment where devices that mediate between high- and low-trust environments are
segregated (from “Demilitarized Zone”)
DNSSEC DNS Security — cryptographically ensures the integrity of DNS information using digital signatures.
DoS
Denial-of-Service (attack)
Dual-stack network
Carries IPv6 and IPv4
Edge firewall
North-South firewall directly connected to Internet
East-West (EW) (Traffic) between portions of the intranet
FIPS
(US) Federal Information Processing Standard (published by NIST)
Firewall Mode
A BIG-IP AFM operating mode in which the implicit policy for every listener (virtual server or self IP)
discards all traffic (so to accept traffic to a listener you must enforce a named policy on it)
Five-Tuple
Basic data from the IP packet header which firewall rules may match: IP protocol, source address and
port, destination address and port. BIG-IP AFM firewall rules can match additional packet attributes
FQDN
A Fully-Qualified Domain Name (FQDN) is a DNS domain name (for example, www.f5.com.) with a Top
Level Domain (TLD) name such as com or org as the rightmost component (rfc1034). Technically an
FQDN (also called an "absolute" domain name) must end with a dot (.) but the trailing dot is almost
always omitted.
HIPAA US regulations for security of health data (from “Health Insurance Portability and Accountability Act”)
HSTS
HTTP Strict Transport Security—a protocol extension to avert crypto downgrade attacks against
secure websites (rfc6797)
F5 Deployment Guide
71
Data Center Firewall
ICAP
A protocol for integrating malware-detecting and DLP tools with network infrastructure devices (from
“Internet Content Adaptation Protocol” rfc3507)
ICSA Labs
Firm which certifies computer security products
IDS/IPS
Intrusion Detection/Prevention System—tools that scan network traffic looking for signatures of
malware and symptoms of malicious activity. Both log suspicious traffic, IPS tries to interrupt it as well
IF-MAP A protocol for interrogating network configuration data, which BIG-IP uses to dynamically associate
user identities with IP addresses (from “Interface for Metadata Access Points”)
Implicit Policy
A dynamic BIG-IP AFM network firewall policy which applies when no named policy has been enforced
on a traffic-handling object
IP Geolocation
Associating a public IP address with a geographic region or country
IP Intelligence
Component of BIG-IP AFM which categorizes remote IP addresses by types of traffic they source or sink
iRuleTM Event-driven script used to extend the capabilities of BIG-IP
ISP
Internet Service Provider
Management Network
A portion of the intranet dedicated to system administration work. Generally a high-trust environment
maintained and guarded with special care
Listener
A logical BIG-IP object which processes network traffic. The most common type of listener is a virtual server
Negative Security
A model for security policy which defines what is forbidden and implicitly permits everything else
Network Firewall
Component of BIG-IP AFM which filters network connections mainly (though not only) at L2–4
NGFW
Next-Generation Firewall—tool combines a zone-based firewall with an IPS and is intended mainly to
control traffic originating from organizational intranet. Like web proxies NGFW’s are used to enforce AUP
NIST
US National Institutes for Standards and Technology (a government agency)
North-South (NS) (Traffic) between Internet and intranet
PCIDSS
Payment Card Industry Data Security Standards. Standards for firms handling card payment data
Pinhole Rule
A positive-model firewall rule to match and admit some narrowly-defined traffic
Positive Security
A model for security policy which defines what is permitted and implicitly forbids everything else
Private (IP) Address
An address meaningful only within an intranet or non-public inter-network. Packets bearing private
addresses cannot be routed over the public Internet
Protocol Security
Component of BIG-IP AFM which enforces compliance with higher-level protocols like HTTP
Proxy Gateway
Device which mediates communications between other network correspondents while maintaining
separate connections to each. By re-encapsulating (and possibly altering) data received on one
connection for transmission on the other, proxy gateways can enable communication at one OSI layer
between correspondents using different protocols at another layer
Public (IP) Address
Packets bearing public addresses may be routed over the public Internet
SIEM
Security Incident/Event Management system (typically accepts and analyzes log messages)
Symbolic Address
A long-lived name (such as an FQDN) for a network resource which may be translated into one or
more short-lived numeric network addresses.
F5 Deployment Guide
72
Data Center Firewall
TMOS
F5 BIG-IP Traffic Management Operating System
TMSH
BIG-IP Traffic Management Shell; command-line interface to TMOS
Trust Environment
A logical portion of the network within which entities communicate freely
VPN
Virtual Private Network—a private network link over shared infrastructure. One commonly assures
privacy by data encryption
Web Proxy
A type of proxy gateway used to control traffic (chiefly web but often some other types as well)
originating from the intranet. Typically enforces AUP and detects malware in data from Internet. Often
integrated with cache (for performance) and DLP
F5 Deployment Guide
73
Data Center Firewall
74
Document Revision History
Version
Description
Date
1.0
New deployment guide
07-06-2015
1.1
dded support for BIG-IP v12.0. The main change is that in version 11.x, an iRule is needed to implement HSTS.
A
BIG-IP v12.0 and later includes support for HSTS (although the iRule will still work). HSTS support in v12 is enabled
through an HTTP profile (detailed in Configuring the BIG-IP system to send web application traffic into the DMZ on
page 37).
09-29-2015
- Added Appendix C: Securing AFM Domain Name Resolution on page 66.
1.2
- Added a link to the F5 NIST SP800-53r4 deployment guide.
02-05-2016
- Added information about symbolic addresses to AFM Network Firewall Rules on page 54
F5 Networks, Inc. 401 Elliott Avenue West, Seattle, WA 98119
F5 Networks, Inc.
Corporate Headquarters
[email protected]
F5 Networks
Asia-Pacific
[email protected]
888-882-4447
F5 Networks Ltd.
Europe/Middle-East/Africa
[email protected]
www.f5.com
F5 Networks
Japan K.K.
[email protected]
©2016 F5 Networks, Inc. All rights reserved. F5, F5 Networks, the F5 logo, and IT agility. Your way., are trademarks of F5 Networks, Inc. in the U.S. and in certain other countries. Other F5 trademarks are identified
at f5.com. Any other products, services, or company names referenced herein may be trademarks of their respective owners with no endorsement or affiliation, express or implied, claimed by F5. 0412
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement