chapter 7: check point qos - Check Point Software Technologies, Ltd.

chapter 7: check point qos - Check Point Software Technologies, Ltd.
.....
C HAPTER 7: C HECK P OINT Q O S
...................................
Quality of Service (QoS) is a set of intelligent network protocols and services
used to efficiently manage the movement of information through local and
wide-area networks. QoS allows Security Administrators to prioritize traffic
flows and provide better service to certain flows. Check Point QoS is included
in every NGX product installation.
Objecti ves
Given a variety of Check Point QoS configurations, determine how to allocate
bandwidth:
1. Determine if Check Point QoS is an appropriate solution, given a variety of
business scenarios.
2. Configure Check Point QoS to meet the requirements, given a variety of
business requirements.
3. Determine how bandwidth will be allocated, given a variety of Check Point
QoS configurations.
4. Identify situations where Low Latency Queueing and Differentiated
Services are an appropriate part of a QoS solution.
325
Chapter 7
7
.....
Key Terms
• Check Point QoS
• IQ Engine
• Weighted Flow Random Early Drop (WFRED)
• Retransmission Detection Early Drop (RDED)
• Differentiated Services (DiffServ)
• Weighted Fair Queuing (WFQ)
• Low Latency Queuing
326
.....
Check Point QoS Overview
CHECK POINT QOS OVERVIEW
..................................................
Check Point QoS is a bandwidth-management solution used to alleviate
bandwidth congestion at network-access points. Check Point QoS controls both
inbound and outbound traffic flows. Check Point QoS uses four technologies to
control traffic:
• Stateful Inspection
• Intelligent Queuing Engine (IQ Engine)
• Weighted Flow Random Early Drop (WFRED)
• Retransmission Detection Early Drop (RDED)
STATEFUL INSPECTION
Check Point QoS uses Check Point’s patented Stateful Inspection technology to
derive complete state and context information for all network traffic.
INTELLIGENT QUEUING ENGINE (IQ ENGINE)
Check Point QoS’s IQ Engine uses state-derived information to classify traffic
and place it in the proper transmission queue. Check Point QoS uses a packet
scheduler to move packets through a dynamically changing scheduling tree at
different rates, in accordance with the QoS Policy.
WEIGHTED FLOW RANDOM EARLY DROP (WFRED)
Weighted Flow Random Early Drop (WFRED) is a mechanism for managing
packet buffers, by selectively dropping packets during periods of network
congestion. WFRED is transparent to users and requires no configuration.
RETRANSMISSION DETECTION EARLY DROP (RDED)
Retransmission Detection Early Drop (RDED) is a mechanism for reducing
the number of retransmits and retransmit storms during periods of network
congestion. RDED prevents inefficiencies by detecting retransmits in TCP
streams and preventing the transmission of redundant packets when multiple
copies of a packet are concurrently queued on the same flow.
327
.....
Check Point QoS Overview
Check Poi nt QoS Archi tecture
The architecture and flow of Check Point QoS is similar to VPN-1 NGX.
Check Point QoS has three components:
SmartConsole — Configures and monitor Check Point QoS
SmartCenter Server — Stores and distribute the Check Point QoS Policy
QoS Module — Enforces Check Point QoS Policy
Like other NGX products, Check Point QoS components can be installed in
stand-alone or distributed deployments.
SMARTCONSOLE AND CHECK POINT QOS
The SmartConsole suite of tools allows a Security Administrator to configure,
verify, install, and monitor the Check Point QoS Policy. SmartDashboard is
used to configure the Policy. SmartView Tracker, SmartView Monitor, and
Eventia Reporter provide useful information that can be used to monitor and
tune the Policy.
The QoS information in SmartView Monitor is provided by
interface. When reviewing Traffic > Top QoS Rules, and
interface must be selected. If an interface is not selected,
SmartView Monitor will display an error message.
Check Point’s integrated architecture allows Administrators to reuse objects
created for Security Policies when defining QoS Policies in SmartDashboard. A
single SmartCenter Server can control and monitor multiple QoS modules.
SMARTCENTER SERVER AND CHECK POINT QOS
After the QoS Policy is defined in SmartDashboard, it is downloaded to the
SmartCenter Server. The SmartCenter Server verifies the Policy and distributes
it to QoS Modules. Policy distribution is handled by the Check Point Daemon
(CPD), which runs on both the SmartCenter Server and QoS Module. The
SmartCenter Server also manages the Check Point Log Repository, and acts as
a log server for SmartView Tracker.
328
.....
Check Point QoS Overview
QOS MODULE
The QoS Module enforces the QoS Policy and provides log information to the
SmartCenter Server. The QoS Module’s functions are divided between the QoS
kernel driver and QoS daemon. The QoS kernel driver examines, queues,
schedules, and releases packets. The QoS daemon is a user-mode process that
performs tasks that are difficult for the kernel. The QoS daemon performs the
following tasks for the kernel:
• DNS name resolution
• Resolving authenticated data for an IP (UserAuthority integration)
• Updating the kernel regarding cluster status (in a Load Sharing
configuration)
The Check Point QoS module must be installed with VPN-1 NGX.
Both products share similar architecture, and Check Point QoS
relies on NGX technology to perform its bandwidth-management
functions.
Check Point QoS Deployment Consi derat ions
A distributed deployment is strongly recommended for Check Point QoS.
Distributed deployments scale gracefully, and dedicated servers reduce the
likelihood that competition for server resources will create a bottleneck. CPU
performance is the main factor affecting the performance of Check Point QoS
modules. A server-class system is strongly suggested. Memory requirements
are based on the number of connections the Check Point QoS Module is
expected to manage. The table below provides guidelines for memory. On
average, each connection requires 1,300 bytes of memory.
Number of Connections
Suggested Memory
10,000
39 MB
25,000
57 MB
50,000
91 MB
100,000
156 MB
329
.....
Check Point QoS Policy
CHECK POINT QOS POLICY
..................................................
Defining a Check Point QoS Policy is very similar to defining a Security Policy.
Rules and objects define the actions a QoS Module should take for different
types of traffic. In the QoS Policy, connections are classified according to four
criteria:
Source — Nodes, networks, user groups, domains
Destination — Nodes, networks, user groups, domains
Resources — Only resources of type URI for QoS
Service — Limited to:
• TCP
• Compound TCP
• UDP
• ICMP
• Citrix TCP Services
• IP Services
Time — Defined for specific times and/or specific days
These fields can be populated with the same objects used to define Security and
Desktop Policies. The Action configuration options for the QoS Policy differ
significantly from the actions available in a Security or Desktop Policy. The
Action field in a QoS rule is used to:
• Define Weights.
• Set Limits.
• Set Guarantees.
Other QoS parameters are configured on the QoS Modules’ Topology >
Interface Properties screens and on the Global Properties > QoS screen.
330
.....
Check Point QoS Policy
Check Point QoS Rul e Base
The Check Point QoS Rule Base can be configured in either Express or
Traditional mode. Express mode allows an Administrator to define basic
Policies quickly and easily. Traditional mode incorporates the more advanced
features of Check Point QoS.
EXPRESS MODE FEATURES
The following Check Point QoS features are available in Express mode:
• Weights
• Per-rule guarantees and limits
• Logging and accounting
• Hardware-accelerator support
• High Availability and Load Sharing
TRADITIONAL MODE FEATURES
In addition to all the features available in Express mode, Traditional mode also
has the following features available:
• Authenticated QoS (UserAuthority integration)
• Per-connection guarantees and limits
• Low Latency Queuing
• Differentiated Services support
• Sub-rules
• Matching by URI resources
• Matching by DNS string
• TCP Retransmission Detection (RDED) mechanism
• Matching Citrix ICA applications
331
.....
Check Point QoS Policy
QoS Acti on Pr oper ti es
The fields in QoS rules can be thought of as defining:
• Who or what:
— Source
— Destination
— Service
• When:
— Time
• How:
— Action
The Action fields in the QoS rules determine how bandwidth will be allocated.
Most actions allow for dynamic allocation of bandwidth, based on current
connections. Low Latency Queuing is an exception, and is discussed later in
this chapter.
SIMPLE AND ADVANCED ACTION TYPES
When an Administrator edits the Action field of a QoS rule, either Simple or
Advanced Action Types may be selected. The configuration options for Simple
Action type allow an Administrator to:
• Apply the rule only to encrypted traffic.
• Set a weight for the rule.
• Set a limit for the rule.
• Set a guarantee for the rule.
The Advanced action type selection adds configuration options for setting perconnection limits and guarantees.
The Advanced action type option is only available for Traditional
mode, not Express mode. If per-connection limits and guarantees
are needed, convert the Policy to Traditional mode, using the
Policy > Convert To option from the main menu.
332
.....
Check Point QoS Policy
WEIGHTS
Weights are used to set the relative priority for different traffic flows. Weight is
the relative portion of available bandwidth allocated to a rule or connection.
Bandwidth is allocated dynamically, and only open connections are included in
the calculation. The formula for determining a rule’s portion of the available
bandwidth is:
this rule’s portion = this rule’s weight / total weight of all rules with open
connections
Weights can only be set per rule, not per connection.
GUARANTEES
A guarantee allocates a minimum amount of bandwidth to the connections
matched with a rule. Guarantees can be defined per rule and per connection.
When guarantees are specified per connection, the Administrator should also
determine how many connections will be permitted. Weights ensure
proportional shares, but only guarantees allow an Administrator to specify an
absolute bandwidth value. Guaranteed bandwidth is allocated before bandwidth
is distributed according to weights.
LIMITS
Limits define the maximum bandwidth that can be assigned to connections
matching a rule. A limit defines a point beyond which connections under a rule
are not allocated bandwidth. A limit can be defined for the entire rule, thus
setting an absolute maximum quantity of bandwidth the rule may consume.
Limits can also be set for each connection matched with a rule.
DEFAULT RULE
A default rule is automatically added to each QoS Policy Rule Base. The weight
of the default rule is defined on the QoS screen in Global Properties. The
default rule applies to all connections not matched by other rules or sub-rules.
The default rule can be modified, but it cannot be deleted.
333
.....
Check Point QoS Policy
Bandwi dth Al locati on and Rules
The following sections provide examples of bandwidth allocation and rules.
Each example includes QoS Rule Base information, open connections,
available bandwidth, and an explanation of how to determine how much
bandwidth a particular connection will be allocated.
Check Point QoS ensures optimal bandwidth use, by allocating
bandwidth dynamically. Each example provided is a “snapshot”.
In a live network, connections open and close constantly. By the
time an Administrator figures out per-connection allocation, the
allocations will have changed.
SIMPLE QOS RULE BASE WITH WEIGHTS ONLY
Adam is the Administrator at ABC Company. ABC Company has only two
networks, Execu_Net and Other_Net. Adam has been told by upper
management that traffic from Execu_Net needs to be given priority over traffic
from Other_Net.
Adam configures a simple QoS Policy with the following rules:
Name
Executive Net Rule
Default
Source
Execu_Net
Any
Destination Any
Any
Service
Any
Any
Action
Weight 30
Weight 10
Track
Log
Log
Install On
Any
Any
Time
Any
Any
Comment
Give Execu_Net
Priority
Blank
The total bandwidth available to the QoS Module is 190 KBps.
334
.....
Check Point QoS Policy
The current open connections on the QoS Module are:
• 8 HTTP connections from Execu_Net.
• 3 SMTP connections from Other_Net.
• 2 RealAudio connections from Execu_Net.
Bandwidth is distributed among the open connections according to weight,
using the formula:
this rule’s portion = this rule’s weight / total weight of all rules with open
connections
In this example, connections match both rules, so the total weight of all rules
with open connections is 40 (Executive Net Rule = 30, Default Rule = 10).
The bandwidth allocated to the Executive Net Rule is:
(30/40) * 190 = 142.5 KBps
The 142.5 KBps allocated to the Executive Net Rule is shared among the 8
HTTP sessions and the 2 RealAudio sessions.
The bandwidth allocated to the Default Rule is:
(10/40) * 190 = 47.5 KBps
The 47.5 KBps is shared among the 3 SMTP sessions from Other_Net.
Q.) How would bandwidth be allocated, if all connections
originated from Execu_Net?
A.) All 190 KBps would be shared among the connections from
Execu_Net. If no open connections match a QoS rule, that rule is
not allocated any bandwidth.
Q.) How would bandwidth be allocated, if no open connections
matched the Executive Net Rule?
A.) All bandwidth would be allocated to the Default Rule.
335
.....
Check Point QoS Policy
QOS RULES BY SOURCE AND SERVICE WITH WEIGHTS ONLY
Bonnie is the Administrator at DEF Corporation. DEF Corporation uses IMAP
as part of their e-mail solution. Bonnie has been instructed to give priority first
to IMAP traffic, and second to the marketing network. All other traffic can be
allocated bandwidth by the Default Rule.
Bonnie configures a QoS Policy with the following rules:
Name
IMAP Rule
Marketing Rule
Default Rule
Source
Any
Marketing_Net
Any
Destination
Any
Any
Any
Service
IMAP
Any
Any
Action
Weight 50
Weight 20
Weight 10
Track
None
None
None
Install On
Any
Any
Any
Time
Any
Any
Any
Comment
Prefer IMAP
Blank
Blank
The total bandwidth available to the QoS Module is 260 KBps.
The current open connections on the QoS Module are:
• 23 IMAP connections from Engineering_Net.
• 32 IMAP connections from Marketing_Net.
• 57 HTTP connections from Marketing_Net.
In this example, only the IMAP and Marketing Rules have connections that
match, so the Default Rule is not used in calculating the total weight of all rules
with open connections. The total weight of all rules with open connections is 70
(IMAP Rule = 50, Marketing Rule = 20).
Traffic is processed on the first QoS rule it matches. In this example, the 32
IMAP connections from Marketing_Net and the 23 IMAP connections from
Engineering_Net all match the IMAP Rule. The 57 HTTP connections from
Marketing_Net match the Marketing Rule.
336
.....
Check Point QoS Policy
The bandwidth allocated to the IMAP Rule is:
(50/70) * 260 KBps = 185.7 KBps
The 185.7 KBps allocated to the IMAP Rule is shared among the 55
connections matching that rule.
The bandwidth allocated to the Marketing rule is:
(20/70) * 260 KBps = 74.3 KBps
The 74.3 KBps allocated to the Marketing Rule is shared among the 57
connections matching the rule.
Q.) How would the bandwidth allocation change if all other
connections stayed open, but an additional connection was opened
from the Engineering_Net using the Telnet service?
A.) The Telnet session from the Engineering_Net would match the
Default Rule. Check Point QoS would dynamically reallocate the
bandwidth to provide the Default Rule its portion of bandwidth.
The total weight of all rules with open connections becomes 80.
Bandwidth is allocated by weight:
(50/80) * 260 KBps = 162.5 KBps for the IMAP Rule
(20/80) * 260 KBps = 65 KBps for the Marketing Rule
(10/80) * 260 KBps = 32.5 KBps for the Default Rule
Q.) How is the bandwidth divided, by connection, in the question
above?
A.) The connections matching each rule share the allocated
bandwidth, so:
55 connections share the 162.5 KBps allocated to the IMAP Rule.
57 connections share the 65 KBps allocated to the Marketing Rule.
1 connection gets the entire 32.5 KBps allocated to the Default
Rule.
337
.....
Check Point QoS Policy
QOS RULES BY SOURCE AND SERVICE WITH WEIGHTS
AND GUARANTEES
Carl is the Administrator at GHI Corporation. Upper management at GHI
Corporation instructed Carl to ensure that 20 percent of the available bandwidth
is guaranteed for RealAudio connections. IMAP connections and connections
from the Executive network should also receive priority. IMAP connections are
a higher priority than connections from the Executive network.
Carl configures a QoS Policy with the following rules:
Name
RealAudio
IMAP
Execu_Net
Default
Source
Any
Any
Execu_Net
Any
Destination
Any
Any
Any
Any
Service
RealAudio
IMAP
Any
Any
Action
Weight 10
Guarantee 100 MBps
Weight 30
Weight 20
Weight 10
Track
None
None
None
None
Install On
Any
Any
Any
Any
Time
Any
Any
Any
Any
Comment
Blank
Blank
Blank
Blank
The total bandwidth available to the QoS Module is 500 MBps.
The current open connections on the QoS Module include the following:
• 7 RealAudio connections from Execu_Net
• 3 RealAudio connections from Marketing_Net
• 298 IMAP connections from Operations_Net
• 11 IMAP connections from Execu_Net
• 22 Telnet connections from Engineering_Net
• 8 IMAP connections from Engineering_Net
• 9 HTTP connections from Execu_Net
• 40 HTTP connections from Operations_Net
• 3 HTTP connections from Engineering_Net
338
.....
Check Point QoS Policy
Bandwidth is allocated to rules with guarantees before it is distributed by
weights. Allocation by guarantee only happens, if there is an open connection
matching the rule. In this example, the RealAudio Rule has a guarantee, and
there are 10 open connections matching that rule. First the QoS Module
allocates 100 MBps to the RealAudio Rule. The remaining 400 MBps is
divided among all rules with open connections, including the RealAudio Rule,
by weight.
All four rules have matching connections, so the total weight of all rules with
open connections is 70. Bandwidth distribution by weight occurs next:
[(10/70) * 400 MBps] + 100 MBps = 157.1 MBps for the RealAudio Rule
(30/70) * 400 MBps = 171.4 MBps for the IMAP Rule
(20/70) * 400 MBps = 114.3 MBps for the Execu_Net Rule
(10/70) * 400 MBps = 57.1 MBps for the Default Rule
Q.) How many connections are sharing the Execu_Net Rule’s
114.3 MBps?
A.) Nine HTTP connections match the Execu_Net Rule. The
RealAudio and IMAP connections from the Execu_Net network
matched on earlier rules.
Q.) If all RealAudio connections are terminated and all other
connections remain open, how is the bandwidth redistributed?
A.) The guarantee is not implemented, and the RealAudio Rule’s
weight is not included in the total weight of all rules with open
connections, thus:
(30/60) * 500 MBps = 250 MBps for the IMAP Rule
(20/60) * 500 MBps = 166.7 MBps for the Execu_Net Rule
(10/60) * 500 MBps = 83.3 MBps for the Default Rule
339
.....
Check Point QoS Policy
QOS RULES BY SERVICE WITH WEIGHTS
AND PER-CONNECTION GUARANTEES
Debbie is the Administrator at JKL Corporation. JKL Corporation uses H.323
for video conferencing. Analysis of the H.323 traffic on her network and a
review of the video-conferencing software have led Debbie to the conclusion
that each H.323 connection requires 384-512 KBps. No more than four H.323
connections will be open concurrently. So Debbie chooses not to allow
additional connections beyond the four permitted in the per connections setting
in the rule’s Action properties. All other connections are of equal priority.
Debbie configures a QoS Policy with the following rules:
Name
Video_Conferencing
Default
Source
Any
Any
Destination
Any
Any
Service
H.323
Any
Action
Weight 1
PC 512 KBps
Weight 10
Track
None
None
Install On
Any
Any
Time
Any
Any
Comment
Blank
Blank
The total bandwidth available to the QoS Module is 8 MBps (8,192 KBps).
If the four H.323 connections are open and there are traffic flows matching the
Default Rule, bandwidth will be allocated first to the per-connection guarantees
and then by weight. Each H.323 connection will be allocated 512 KBps, and the
remaining bandwidth will be distributed by weight, across all rules, including
the H.323 Rule.
8,192 - (4 * 512) = 6,144 KBps to distribute by weight
[(1/11) * 6,144] + (4 * 512) = 2,662.4 KBps allocated to the H.323 Rule
(9/11) * 6,144 = 5,529.6 KBps shared by all flows matching the Default Rule
340
.....
Check Point QoS Policy
Q.) If all four H.323 connections remained open, and the traffic
flows matching the Default Rule all closed, how would bandwidth
be allocated?
A.) The remaining 6,144 KBps would be shared among the four
connections matching the H.323 Rule.
Q.) Since Debbie did not allow additional connections in the
Action properties, what happens if someone attempts to establish a
fifth H.323 connection?
A.) No bandwidth will be allocated to the fifth H.323 connection.
If Debbie checks Accept Additional Connections in the H.323 Rule’s Action
properties, bandwidth will be allocated to additional connections by weight. If
four H.323 connections have guaranteed bandwidth of 512 KBps, and two
additional connections are opened and connections match the Default Rule,
bandwidth will be allocated:
1. The four guaranteed H.323 connections will be given their 512 KBps.
2. The Default Rule will receive its 10/11 of the remaining bandwidth (5,529.6
KBps).
3. The remaining 1/11 of the bandwidth (614.4 KBps) will be shared among
the six H.323 connections:
614.4 KBps for the first 4 H.323 connections and
102.4 KBps for the next 2 H.323 connections
Since Debbie’s research indicates 384-512 KBps are required to sustain
minimum quality for the H.323 sessions, she should consider either increasing
the weight of the H.323 Rule or not allowing additional connections.
The sum of guarantees in rules in the upper level should not
exceed 90 percent of the capacity of the link.
341
.....
Check Point QoS Policy
QOS BY SOURCE AND SERVICE WITH LIMITS AND WEIGHTS
Edward, the Administrator for MNO Corporation, must ensure that the total
bandwidth consumed by FTP traffic never exceeds 250 KBps. Also, traffic
from Execu_Net should receive higher priority than other traffic.
Edward configures a QoS Policy with the following rules:
Name
FTP_Limit
Execu_Net
Default
Source
Any
Execu_Net
Any
Destination
Any
Any
Any
Service
FTP
Any
Any
Action
Weight 10
Limit 250 KBps
Weight 20
Weight 10
Track
None
None
None
Install On
Any
Any
Any
Time
Any
Any
Any
Comment
Blank
Blank
Blank
The total bandwidth available to the QoS Module is 2 MBps (2,048 KBps).
If connections match all three rules in the QoS Policy, bandwidth will be
allocated, as follows:
250 KBps for the FTP_Limit Rule, because its allocation cannot exceed the
defined limit
(20/30) * 1,798 = 1,198.7 KBps for the Execu_Net Rule
(10/30) * 1,798 = 599.3 KBps for the Default Rule
If the bandwidth allocated by weight is less than the limit, the
rule will receive the lower allocation. In the example above, if
Edward’s QoS Module had only 512 KBps total available
bandwidth, and all three rules had open connections matching
them, the FTP_Limit Rule would have been allocated 128 KBps.
342
.....
Check Point QoS Policy
QOS BY SOURCE WITH WEIGHTS AND PER-CONNECTION LIMITS
Edward analyzes the traffic at MNO Corporation, and determines that limiting
each FTP connection to no more than 20 KBps will improve his compliance
directives from management. He changes the FTP_Limit Rule to read as
follows:
Name
FTP_Limit
Source
Any
Destination
Any
Service
FTP
Action
Weight 10
Limit 250 KBps
LC 20 KBps
Track
None
Install On
Any
Time
Any
Comment
Blank
Now any single FTP connection will be able to consume no more than 20
KBps, and the sum total of all FTP connections will be able to consume no
more than 250 KBps.
When per-connection limits are imposed, the number of connections matching
the rule with per-connection limits has an impact on the total bandwidth
remaining for allocation by weight. For example, if only two FTP connections
were open, then only 40 KBps would be allocated to the FTP_Limit Rule. The
remaining 2,008 KBps would be distributed among connections matching the
Execu_Net and Default Rules as follows.
343
.....
Check Point QoS Policy
GUARANTEE AND LIMIT INTERACTION
If a rule limit and guarantee per rule are defined in a rule, the limit should not
be smaller than the guarantee. A combination of guarantees and limits can be
used instead of very low weights, to ensure connections get the bandwidth they
need, but do not consume a large amount of the bandwidth allocated in the
weight distribution.
Recall the example with Debbie and JKL Corporation. Debbie’s research
indicated that each H.323 connection needed from 384-512 KBps to sustain
optimal quality. Debbie could use a per-connection guarantee to make sure each
H.323 connection received at least 384 KBps, and a per-connection limit to
make sure none received more than 512 KBps.
Debbie configures a QoS Policy with the following rules:
Name
Video_Conferencing
Default
Source
Any
Any
Destination
Any
Any
Service
H.323
Any
Action
Weight 10
PC 384 KBps
LC 512 KBps
Weight 10
Track
None
None
Install On
Any
Any
Time
Any
Any
Comment
Blank
Blank
The total bandwidth available to the QoS Module is 8 MBps (8,192 KBps).
344
.....
Check Point QoS Policy
Debbie only provides the per-connection guarantee for four connections, but
she allows additional connections in the Action properties. Notice that she also
changed the weight for the Video_Conferencing Rule from 1 to 10, so it is now
equal to the weight for the Default Rule. If there are connections matching both
the Video_Conferencing Rule and the Default Rule, and there are six H.323
sessions open, bandwidth will be allocated as follows:
1. 384 KBps to each of the first four H.323 sessions (total of 1,536 KBps)
2. All six H.323 connections will be taken up to 512 KBps, provided it does
not exceed the bandwidth allocation by weight. In this case, the by-weight
allocation would have been 3,328 KBps, but only 1,536 KBps is required to
take each connection to its 512 KBps limit.
3. The remaining 5,120 KBps is allocated to open connections matching the
Default Rule.
BANDWIDTH ALLOCATION AND SUB-RULES
Sub-rules are rules within a rule. Sub-rules allow an Administrator even more
granular control over bandwidth allocation, but permit weights, guarantees, and
limits within top-level rules. Sub-rules are only available in Traditional mode in
the Rule Base.
Add Sub-Rule — QoS only, to add sub-rules to QoS rules
Add QoS Class — Adds a QoS Class of Service to a QoS rule
When a sub-rule is created for a top-level rule, a default rule for
the sub-rule set is also generated. The default rule for the sub-rule
set determines the fate of traffic that does not match any of the
defined sub-rules in the set.
George is the Administrator at PQR Corporation. PQR Corporation provides a
wide variety of services to its customers, via Web applications over HTTP.
George needs to ensure that HTTP traffic receives higher priority than all other
traffic, but he also must also make sure that inbound HTTP traffic receives
higher priority than outbound HTTP traffic.
345
.....
Check Point QoS Policy
George configures a QoS Policy with the following rules:
Name
Web_Traffic
Default
Source
Any
Any
Destination
Any
Any
Service
HTTP
Any
Action
Weight 20
Weight 10
Track
None
None
Install On
Any
Any
Time
Any
Any
Comment
Blank
Blank
The total bandwidth available to the QoS Module is 8 MBps (8,192 KBps).
George creates a network object “PQR_Net”, to represent the supernet that
encompasses all of PQR’s internal networks. George then configures two subrules for the Web_Traffic Rule. (The Default Rule for the Web_Traffic sub-rule
set was added automatically by Check Point QoS, when the first sub-rule was
created.)
Name
Inbound_Web
Outbound_Web
Default
Source
Any
PQR_Net
Any
Destination
PQR_Net
Any
Any
Service
Any
Any
Any
Action
Weight 50
Weight 20
Weight 10
Track
None
None
None
Install On
Any
Any
Any
Time
Any
Any
Any
Comment
Blank
Blank
Blank
There is no need to define the service for these sub-rules, as only
HTTP traffic with be passed down from the top-level rule. The
Source, Destination, and Service fields of a sub-rule must always
be a subset of the parent rule, or the sub-rule will not be effective.
346
.....
Check Point QoS Policy
Connections match both the top-level rules (Web_Traffic and Default), so
bandwidth is allocated between the two rules according to weight:
(20/30) * 8,192 = 5,461.3 KBps for the Web_Traffic Rule
(10/30) * 8,192 = 2,730.7 KBps for the Default Rule
Connections match all three sub-rules of the Web_Traffic Rule. The bandwidth
allocated to the Web_Traffic Rule is further subdivided, according to the
weights of the sub-rules:
(50/80) * 5,461.3 = 3,413.3 KBps for the Inbound_Web sub-rule
(20/80) * 5,461.3 = 1,365.3 KBps for the Outbound_Web sub-rule
(10/80) * 5,461.3 = 682.7 KBps for the Default sub-rule
Guarantees and limits within sub-rule sets are handled the same way as with
top-level rules. The bandwidth is limited to that allocated to the sub-rule set’s
top-level rule.
ADDITIONAL QOS RULE CONSIDERATIONS
Some additional considerations for QoS rules include:
• If a guarantee is defined in a sub-rule, a guarantee must be defined for the
rule above it.
• The guarantee of a sub-rule cannot be greater than the guarantee defined for
the rule above it.
• A rule guarantee must not be smaller than the sum of guarantees defined in
its sub-rules.
• If a rule’s weight is low, some connections may receive very little
bandwidth.
• The sum of guarantees in rules in the upper level should not exceed 90
percent of the capacity of the link.
• If per-connection guarantees are defined both for a rule and its sub-rules, the
sum of the per-connection guarantees for the sub-rules should not be greater
than the per-connection guarantee of the top-level rule.
347
.....
Check Point QoS Policy
• If both a rule and per-connection limit are defined for a rule, the perconnection limit must not be greater than the rule limit.
• If a limit is defined in a rule with sub-rules and limits are defined in all subrules, the rule limit should not be greater than the sum of limits defined in
the sub-rules.
• If a rule limit and a guarantee per rule are defined in a rule, the limit should
not be smaller than the guarantee.
If both a limit and a guarantee are defined in a rule and the limit is
equal to the guarantee, connections may receive no bandwidth.
This situation can occur when the rule has sub-rules with total rule
guarantees that add up to the total rule guarantee for the entire rule,
and the rule also has sub-rules with no guarantee.
348
.....
Differentiated Services
DIFFERENTIATED SERVICES
..................................................
Differentiated Services (DiffServ) is an architecture for providing different
types or levels of service for network traffic. Inside the enterprise network,
packets are marked in the IP header TOS byte as belonging to a certain class of
service, or QoS class. These packets are then granted priority on the public
network. DiffServ packets have meaning on the public network, not inside the
enterprise network. Effective implementation of DiffServ requires that packet
markings be recognized on all public network segments.
Di ff Ser v Mar ks for I PSec Packet s
When DiffServ marks are used for IPSec packets, the DiffServ mark can be
copied from one location to another in one of two ways:
:ipsec.copy_TOS_to_inner
The DiffServ mark is copied from the IPSec header to the IP header of the
original packet, after decapsulation/decryption:
:ipsec.copy_TOS_to_outer
The DiffServ mark is copied from the original packet’s IP header to the IPSec
header of the encrypted packet, after encapsulation.
This property should be set, per QoS Module, in $FWDIR/conf/objects_5_0.c.
The default setting is:
:ipsec.copy_TOS_to_inner (false)
:ipsec.copy_TOS_to_outer (true)
349
.....
Differentiated Services
Int eracti on between Di ff Serv Rul es and Other Rules
A DiffServ rule specifies not only a QoS class, but also a weight, in the same
way that other QoS Policy rules do. These weights are enforced only on the
interfaces on which the rules of this class are installed.
For example, suppose a DiffServ rule specifies a weight of 50 for FTP
connections. That rule is installed only on the interfaces for which the QoS
class is defined.
Interface QoS Properties Tab
On other interfaces, the rule is not installed and FTP connections routed
through those other interfaces do not receive the weight specified in the rule. To
specify a weight for all FTP connections, add a rule under “Best Effort.”
DiffServ rules can be installed only on interfaces for which the relevant QoS
class has been defined in the QoS tab of the Interface Properties screen. “Best
Effort” rules (that is, non-DiffServ rules) can be installed on all interfaces of
Gateways with QoS Modules installed. Only rules installed on the same
interface interact with each other.
350
.....
Low Latency Queuing
LOW LATENCY QUEUING
..................................................
For most traffic, Weighted Fair Queuing (WFQ) is adequate. WFQ helps
avoid drops caused by congestion. Avoiding drops can means holding long
queues. Long queues may lead to non-negligible delays.
Long queues are inappropriate for voice and video traffic. For most delaysensitive applications, packets need not be dropped from queues to keep them
short. The streams of these applications have a known, bounded bit rate. Check
Point QoS can be configured to forward as much traffic as the stream delivers,
ensuring only a small number of packets accumulate in the queues.
Low Latency Queuing allows an Administrator to define special classes of
service for delay-sensitive applications. Rules under these classes are used with
other rules in the QoS Rule Base.
For each Low Latency class defined on an interface, specify a constant bit rate
and maximal delay for active directions. Check Point QoS checks packets
matched to Low Latency class rules, to prevent them from being delayed longer
than their maximal delay. If the maximal delay of a packet is exceeded, the
packet is dropped. Otherwise, packets are transmitted at the constant bit rate
defined for the Low Latency class to which it belongs. If the constant bit rate of
the class is defined correctly (meaning that it is not smaller than the expected
arrival rate of the matched traffic), packets are not dropped. When the arrival
rate is higher than the specified constant bit rate, packets exceeding this
constant rate are dropped, to ensure that those transmitted are within the
maximal delay limitations.
Low Lat ency Cl asses
Low Latency classes specify the maximal delay that is tolerated and a constant
bit rate. Check Point QoS guarantees traffic matching rules of this type is
forwarded within the limits of the bounded delay.
351
.....
Low Latency Queuing
DEFINING A LOW LATENCY CLASS
To define a Low Latency Class:
1. Select Manage > QoS > QoS Classes from SmartDashboard menu.
2. Click the New button, and select Low Latency Class of Service.
3. Give the class a name, and assign it a priority.
Low Latency Class Properties
4. Select the target QoS Module. Go to the Topology screen, and select the
interface to edit.
5. Click the Add button, and select Low Latency Classes.
6. Populate the Add Low Latency QoS Class Properties screen, and click OK
to exit the properties screens.
Low Latency QoS Class Properties
7. Right-click the QoS Rule Base, and select Add Class of Service > Above.
352
.....
Low Latency Queuing
8. Select the target class of service from the Add Class of Service screen, and
click OK. The Class of Service Rule is added to the QoS Rule Base. A Best
Effort Rule is also added:
Low Latency Class Rule and Best Effort Rule
9. To activate the Low Latency Class, define at least one rule under it in the
QoS Policy Rule Base:
Low Latency Class Rule with QoS Policy Rule
The traffic matching any Low Latency Class rule receives the delay and
constant bit rate properties defined for the specified class, and is also
allocated bandwidth according to the rule properties (weight, limit, and
guarantee).
The maximal delay is an upper limit. Packets matching the class
are always forwarded with a delay not greater, but often smaller,
than specified.
353
.....
Low Latency Queuing
Low Latency Class Priorit ies
It is advisable to define more than one Low Latency Class, if different types of
traffic require different maximal delays. Low Latency Classes are assigned one
of five priority levels. These priority levels are relative to other Low Latency
Classes. The class with the lower maximal delay should get a higher priority
than the class with the higher delay. When two packets are ready to be
forwarded, one for each Low Latency Class, the packet from the higher priority
class is forwarded first. The remaining packet, from the lower class, encounters
greater delay.
The maximal delay that can be set for a Low Latency Class depends on the Low
Latency Classes of higher priority. Other Low Latency Classes can affect the
delay incurred by a class, and must be taken into consideration when setting the
minimal delay for the class. Set the priorities for all Low Latency Classes
according to maximal delay, then define the classes by descending priority.
For each direction of an interface (inbound and outbound), the sum
of the constant bit rates of all the Low Latency Classes cannot
exceed 20 percent of the total designated bandwidth rate. The 20percent limit is set to ensure that Best Effort traffic does not suffer
substantial delay and jitter, as a result of the Low Latency Classes.
COMPUTING CONSTANT BIT RATE
To accurately set the constant bit rate of a Low Latency Class, an Administrator
must know:
• The bit rate of a single application stream of traffic matching the class.
• The expected number of simultaneously open streams.
The constant bit rate of the class should be the bit rate of a single application,
multiplied by the expected number of simultaneous streams:
(Single Stream) * (Number of Streams Expected) = constant bit rate
If the number of streams exceeds the number expected when
setting the constant bit rate, the total incoming bit rate will exceed
the constant bit rate, and many drops will occur. Avoid this
situation, by limiting the number of concurrent streams.
354
.....
Low Latency Queuing
COMPUTING MAXIMAL DELAY
The delay value defined for a class determines the number of packets that can
be queued in the Low Latency queue, before drops begin. When computing the
maximal delay of a Low Latency class, consider both the maximal delay that
streams matching the class can tolerate, and the minimal delay that Check Point
QoS can guarantee.
Use the following method to estimate the greatest delay the class can tolerate:
1. Refer to the technical details of the streaming application, and find the
delay it can tolerate.
2. Find or estimate the delay the external network imposes.
3. Subtract the delay imposed by the external network from the maximum
delay the streaming application can tolerate.
Use the following method to estimate the smallest delay for the class:
1. Find the bit rate of the streaming application.
If you cannot find the bit rate of the streaming application in the
application properties, you can use SmartView Monitor to
observe the application and determine the bit rate.
2. Estimate the typical packet size in the stream.
If you do not know the packet size, you can use fw monitor to
capture traffic and record packet size, or you can use the LAN’s
MTU (1500 bytes for Ethernet).
3. Estimate burst size, by monitoring the internal interfaces that precede the
QoS Module.
If no burstiness is detected, the minimum delay of the class should be no
smaller than this:
(3 * packet size) / bit rate
355
.....
Low Latency Queuing
If burstiness is detected, the minimum delay of the class should be no
smaller than this:
[(burst size + 1) * packet size] / bit rate
4. The maximal delay for a class should be between the greatest and smallest
delay. Set the delay close to the greatest value, only if the application
occasionally bursts.
When you set the maximal delay, you will see an error message, if
the value is lower than the value set for other Low Latency Classes
of higher priority. Check Point QoS will also display an error
message, if the maximal delay is incompatible with interface
speed. Both error messages include a minimum acceptable
maximal-delay value. Set the maximal delay to a value no smaller
than the one printed in the message.
PREVENTING UNWANTED DROPS
If the aggregate bit rate going through the Low Latency Class exceeds the
constant bit rate of the class, drops occur. This situation may occur when the
number of streams actually opened exceeds the number you expected, when
you set the constant bit rate.
Limit the number of connections allowed to the number of connections used to
compute the class’s constant bit rate, by modifying the QoS Policy as follows:
1. Define a single rule under the class, with a per-connection guarantee as its
Action.
2. In the Per Connection Guarantee field of the QoS Action Properties screen,
define the per-connection bit rate you expect.
3. In the Number of guaranteed connections field, define the maximum
number of connections allowed in this class.
4. Do not select Accept additional non-guaranteed connections.
When to Use Low Latency Queuei ng
Use Low Latency Queueing when the bit rate of the stream is known and
controlling delay is important. Low Latency Classes do not receive TOS
markers. If preferential treatment is required beyond the QoS Module, DiffServ
should be used instead.
356
.....
Advanced Features
ADVANCED FEATURES
..................................................
Check Point QoS includes advanced features to allow authenticated QoS,
support Citrix MetaFrame, and integrate with Load Sharing configurations.
Authenti cated QoS
Check Point Authenticated QoS provides Quality of Service for end-users in
dynamic IP environments, such as remote-access and DHCP environments.
Authenticated QoS enables priority users, such as corporate CEOs, to receive
priority service when remotely connecting to corporate resources.
Authenticated QoS dynamically prioritizes end users, based on information
gathered during network or VPN authentication. The feature leverages Check
Point UserAuthority technology, to classify both inbound and outbound user
connections. Check Point QoS supports Client Authentication, including
encrypted Client Authentication, and VPN-1 SecuRemote/SecureClient
authentication. User and Session Authentication are not supported.
Ci tr ix Met aFrame Support
Citrix MetaFrame is a client/server software application that enables a client to
run a published application on a Citrix server farm, from the client's desktop.
One of the disadvantages of using Citrix ICA is that uncontrolled printing
traffic can consume all available bandwidth, leaving mission-critical
applications struggling.
Check Point QoS solves the problem by:
• Classifying all ICA applications running over Citrix through layer 7.
• Differentiating between the Citrix traffic based on ICA published
applications, ICA printing traffic (Priority Tagging), and NFuse.
357
.....
Advanced Features
LIMITATIONS
The following limitations apply when using Check Point QoS to control Citrix
MetaFrame traffic:
• The Citrix TCP services are supported in Traditional mode QoS Policies
only.
• Session Sharing must be disabled.
• The number of applications that are detected by the inspection infrastructure
is limited to 2,048. Console errors will be sent, if this limit is exceeded.
These errors are harmless, and will not affect your system. Simply restart the
machine.
• Versions of MetaFrame prior to 1.8 are not supported, because there is no
packet tagging in these versions.
• Only one Citrix TCP service can be allocated per single rule.
Load Sharing
Load Sharing is a mechanism that distributes traffic within a cluster of
Gateways, so the total throughput of multiple machines is increased. Check
Point QoS’s architecture guarantees that Load Sharing will provide either:
• Two-way stickiness. All packets of a single connection use the same
machine in both directions.
• Conversation stickiness. All packets of control/data connections within a
conversation use the same machine in both directions.
Check Point QoS provides a fault-tolerant QoS solution for cluster Load
Sharing that deploys a unique, distributed WFQ bandwidth-management
technology. The user is able to specify a unified QoS Policy per virtual interface
of the cluster. The resulting bandwidth allocation is therefore identical to that
obtained by installing the same QoS Policy on a single server.
Under a load state, there are a few connections that are
backlogged actively for short periods of time. In such cases, the
Load Sharing function in ClusterXL is not spread evenly. But in
this case, there is no congestion and therefore no need for QoS.
358
.....
Monitoring QoS Policy
MONITORING QOS POLICY
..................................................
The QoS Policy and its impact on traffic can be monitored using SmartView
Tracker, Eventia Reporter, and SmartView Monitor. Each SmartConsole tool
provides a rich set of data, to evaluate the effectiveness of the QoS Policy.
SmartVi ew Tracker
The following two conditions must be met, for a match on a QoS Policy rule to
be logged:
• The QoS logging box must be checked on the Gateway Properties >
Additional Logging Configuration screen.
• The connection’s matching rule must be marked with either Log or Account,
in the rule’s track field.
QOS TRACK FIELD SET TO LOG
QoS rules with the track field set to Log can generate the following types of log
events:
• Connection Rejection
QoS rejects a connection when the number of guaranteed connections is
exceeded, and/or when the rule’s action properties are not set to accept
additional connections.
• Running Out of Packet Buffers
QoS generates an Out of Packet Buffers string when QoS global packet
buffers are exhausted, or one of the interface direction’s packet buffers is
exhausted. This report is generated no more frequently than once every 12
hours.
• LLQ Packet Drop
When a packet is dropped from an LLQ connection, a report is generated.
This report is generated no more frequently than once every 5 minutes.
359
.....
Monitoring QoS Policy
QOS TRACK FIELD SET TO ACCOUNT
QoS rules set to Account generate the following types of events:
• Total bytes transmitted through QoS for each relevant interface and direction
• Total bytes dropped from the connection as a result of the QoS drop Policy
• Count of bytes dropped from the connection, because the maximum used
memory fragments for a single connection was exceeded
• Number of bytes dropped from the connection, due to delay expiration for
LLQ connections
• Average packet delay for LLQ connections
• Jitter (maximum delay difference between two consecutive packets) for
LLQ connections
Smart View Moni tor
SmartView Monitor allows an Administrator to view the top QoS rules by
interface. Information provided in SmartView Monitor can help an
Administrator optimize the QoS Rule Base, by showing which rules should be
placed near the top:
Sample of SmartView Monitor QoS Information
360
.....
Monitoring QoS Policy
Eventi a Reporter
The Eventia Reporter Network Activity report presents data about traffic
accepted by the Gateway. This report can be used to see network activity, to
determine effective allocation of bandwidth.
Specific sections include information regarding:
• Overall traffic characteristics, as well as a breakdown by hour and by date.
• The top network users.
• Top services used.
• Top sources and destinations of network traffic
If an Administrator is concerned that an abundance of HTTP and RealAudio
traffic is consuming excessive bandwidth and starving IMAP connections, the
Top Network Activity - Services section of the Network Activity report
determines if proper application of QoS could improve bandwidth allocation:
Top Network Activity - Services Report
361
.....
Optimizing Check Point QoS
OPTIMIZING CHECK POINT QOS
..................................................
Check Point QoS performance can be improved by following the suggestions
below:
• Upgrade to the newest Check Point QoS version available.
• Install Check Point QoS only on the external interfaces of the QoS Module.
Unless you are using limits for inbound traffic, installing Check Point QoS
only in the outbound direction will provide you the most functionality and
improvements.
• Put more frequent rules at the top of your Rule Base. You can use
SmartView Monitor to analyze how much a rule is used.
• Turn per-connection limits into per-rule limits.
• Turn per-connection guarantees into per-rule guarantees.
362
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