iStack Technology White Paper

iStack Technology White Paper
Issue
01
Date
2013-04-09
HUAWEI TECHNOLOGIES CO., LTD.
Copyright © Huawei Technologies Co., Ltd. 2013. All rights reserved.
No part of this document may be reproduced or transmitted in any form or by any means without prior
written consent of Huawei Technologies Co., Ltd.
Trademarks and Permissions
and other Huawei trademarks are trademarks of Huawei Technologies Co., Ltd.
All other trademarks and trade names mentioned in this document are the property of their respective
holders.
Notice
The purchased products, services and features are stipulated by the contract made between Huawei and
the customer. All or part of the products, services and features described in this document may not be
within the purchase scope or the usage scope. Unless otherwise specified in the contract, all statements,
information, and recommendations in this document are provided "AS IS" without warranties, guarantees or
representations of any kind, either express or implied.
The information in this document is subject to change without notice. Every effort has been made in the
preparation of this document to ensure accuracy of the contents, but all statements, information, and
recommendations in this document do not constitute a warranty of any kind, express or implied.
Huawei Technologies Co., Ltd.
Address:
Huawei Industrial Base
Bantian, Longgang
Shenzhen 518129
People's Republic of China
Website:
http://enterprise.huawei.com
Email:
ChinaEnterprise_TAC@huawei.com
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
i
iStack Technology White Paper
Contents
Contents
1 iStack ............................................................................................................................................... 1
1.1 iStack Overview ............................................................................................................................................... 1
1.2 Advantages of iStack Technology .................................................................................................................... 2
1.3 Principles .......................................................................................................................................................... 5
1.3.1 Concepts.................................................................................................................................................. 5
1.3.2 Principles ................................................................................................................................................ 7
1.4 Applications.................................................................................................................................................... 11
1.5 Example for Configuring the iStack Function ................................................................................................ 12
A A Terms, Acronyms, and Abbreviations ............................................................................... 16
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
ii
iStack Technology White Paper
1 iStack
1
iStack
About This Chapter
1.1 iStack Overview
1.2 Advantages of iStack Technology
1.3 Principles
1.4 Applications
1.5 Example for Configuring the iStack Function
1.1 iStack Overview
Currently, two types of communication devices are available on the network: box devices and
chassis devices.

Box devices are cost-effective, but do not support high availability and uninterrupted
service protection. So box devices cannot be used at the core layer, at the aggregation
layer, or in data centers. In complex networking environment, because box devices have
low scalability, you have to maintain more network devices and change the original
networking structure when adding new devices.

Chassis devices have advantages such as high availability, high performance, and high
port density. Therefore, chassis devices are often used at the core layer, at the
aggregation layer, and in data centers. Compared with box devices, chassis devices have
disadvantages such as a high initial investment and high single port cost.
The iStack technology combines the advantages of both box devices and chassis devices. The
iStack technology virtualizes multiple devices supporting the stacking function into one
logical device as shown in Figure 1-1. This virtual device has advantages like
cost-effectiveness of box devices and high scalability and reliability of chassis devices.
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
1
iStack Technology White Paper
1 iStack
Figure 1-1 Setting up a stack
iStack
iStack Link
iStack is a virtualization technology that virtualizes multiple devices at the same layer into a
logical device without changing the network physical topology structure. The iStack
technology simplifies network structure and network protocol deployment, and improves
network reliability and manageability as shown in Figure 1-2.
Figure 1-2 Network horizontal virtualization
Network
Network
CSS
Eth-Trunk
iStack
Eth-Trunk
iStack
1.2 Advantages of iStack Technology
Simplified Configuration and Management
After a stack is set up, multiple physical devices are virtualized into one logical device. You
can log in to the stack through any member device to configure and manage all the member
devices.
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
2
iStack Technology White Paper
1 iStack
1:1 Redundancy of Control Planes
Chassis switches use the 1:1 redundancy mode. That is, each chassis switch is equipped with
two Main Processing Units (MPUs). The master MPU processes services, and the standby
MPU functions as a backup of the master MPU and synchronizes information with the master
MPU. When the master MPU fails, the standby MPU becomes the new master MPU and
starts to process services.
A box switch has only one control plane and cannot implement redundancy. When a box fails,
the connected network is interrupted. iStack technology can implement 1:1 redundancy on
box switches. In a stack, one master switch processes services, and one standby switch
functions as a backup of the master switch and synchronizes information with the master
switch. The other switches in the stack are slave switches. When the master switch fails, the
standby switch becomes the new master switch, and a new standby switch is selected from the
slave switches.
As shown in Figure 1-3, configuration and data on the standby switch is completed
synchronized with the master switch. Therefore, when the standby switch becomes the new
master switch, it can immediately replace the original master switch to manage other switches
in the stack. Multiple slave switches in the stack further improve system reliability.
Figure 1-3 Standby switch becomes the new master switch
Master
Standby
Slave
Slave
Slave
Master
Standby
Uplink and Downlink Redundancy
iStack can implement redundancy of uplinks and downlinks through inter-device link
aggregation. Traditional link aggregation technology combines multiple physical Ethernet
interfaces (member interfaces) into one logical interface to provide backup when a link fails.
However, this technology cannot provide backup when a device fails.
iStack supports inter-device link aggregation, which allows you to aggregate physical
Ethernet interfaces on multiple member switches of a stack into one logical interface. When a
device some member interfaces fails, the other member switches can manage and maintain the
remaining member interfaces so that services are not interrupted. Inter-device link aggregation
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
3
iStack Technology White Paper
1 iStack
prevents service interruption caused by single-point failures and greatly improves network
availability.
As shown in Figure 1-4, traffic sent to core devices of the network is evenly distributed to
multiple links in a link aggregation group. When a link fails, traffic on this link is evenly
distributed to the other links. This link redundancy mechanism improves network reliability.
Figure 1-4 Data traffic forwarding after a link failure
Flow
Flow
Flow
Eth-Trunk
Eth-Trunk
Master
Slave
Standby
Slave
Master
Standby
Slave
Slave
Redundancy of Stack Links and Stack Interfaces

Redundancy of stack links in a ring topology
iStack can implement redundancy of stack links in a ring topology. When a link fails, the
ring topology changes into a chain topology so that services in the stack are not affected.

Redundancy of stack interfaces
iStack uses link aggregation to implement redundancy of stack interfaces. Multiple
physical links on stack interfaces can be aggregated to load balance traffic, improve
bandwidth and system performance. Additionally, the physical links back up each other
so that the failure of one link does not affect services in the stack. This improves device
reliability.
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
4
iStack Technology White Paper
1 iStack
Figure 1-5 Redundancy of stack links and stack interfaces
Simplified Networking
As shown in Figure 1-6, multiple devices on the aggregation layer are virtualized into a
logical device through iStack technology. This simplified network does not require the
Multiple Spanning Tree Protocol (MSTP) or Virtual Router Redundancy Protocol (VRRP), so
network configuration is much simpler. Inter-device link aggregation also speeds up network
convergence and improves network reliability.
Figure 1-6 Simplified network
MSTP + VRRP
iStack
1.3 Principles
1.3.1 Concepts

Switch roles
Each switch in a stack is a member switch. Member switches are classified into the
following roles:
−
Master switch
The master switch manages the entire stack. A stack has only one master switch.
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
5
iStack Technology White Paper
−
1 iStack
Standby switch
The standby switch is the backup to the master switch. A stack has only one standby
switch.
−
Slave switch
In a stack, all member switches except the master switch are slave switches. The
standby switch is also a slave switch.

Stack domain
Switches that connect to each other using stack links to form a stack belong to a stack
domain. To meet various networking requirements, you can configure multiple stacks on
a network and use stack domain IDs to identify these stacks as shown in Figure 1-7.
Figure 1-7 Multiple stack domains
Domain 1
iStack Link
Domain 2
Domain 3
iStack Link
iStack Link

Stack ID
A stack ID, also called a member ID, is used to identify and manage member switches in
a stack. All member switches in a stack have a unique stack ID.

Stack priority
The stack priority is an attribute of member switches, which helps determine the role of
member switches in role election. A larger priority value indicates a higher priority. The
member switch with a higher stack priority has a higher probability of becoming the
master switch.
Issue 01

Physical member interface
Switches connect to each other to form a stack using physical member interfaces.
Physical member interfaces forward service packets and stack protocol packets between
member switches.

Stack interface
A stack interface is a logical interface that is bound to physical member interfaces to
implement the stacking function. Each member switch has two stack interfaces, which
are named Stack-Portn/1 and Stack-Portn/2. n specifies the stack ID of the member
switch.
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
6
iStack Technology White Paper
1 iStack
1.3.2 Principles
Setting Up a Stack
To use multiple switches to form a stack, connect physical member interfaces bound to the
local stack interface to those bound to the neighbor stack interface, as shown in Figure 1-8. A
stack interface can be bound to multiple physical member interfaces to improve stack link
bandwidth and reliability.
Figure 1-8 Stack networking diagram
Stack
interface
iStack
Stack
interface
iStack Link
The stack contains multiple member switches, each of which has a role. During the setup of a
stack, member switches exchange packets to elect the master switch that manages the stack,
and the other switches become slave switches.
The rules for electing the master switch are as follows:
1.
The switch that has started is preferred over the switch that is starting.
2.
The switch with higher stack priority is preferred.
3.
The switch with a later software version is preferred.
4.
The switch with a smaller MAC address is preferred.
The election results are compared one by one. In the case of the same election result, the next
rule is used until the master switch is elected.
The master switch collects member information, calculates the stack topology, and
synchronizes the stack topology to all the other member switches.

If slave switches have the same stack ID, the master switch assigns a unique stack ID to
each of the member switches.

When the master and slave switches use different software versions, slave switches
synchronize the software version with the master switch, restart, and then join the stack.
The master switch elects a standby switch from slave switches as the standby switch. When
the master switch fails, the standby switch takes over all services from the master switch.
The rules for electing the standby switch are as follows:
1.
The switch with higher stack priority is preferred.
2.
The switch with a smaller MAC address is preferred.
The election results are compared one by one. In the case of the same election result, the next
rule is used until the master switch is elected.
Before a stack is set up, each switch is an independent entity and has its own IP address. You
need to manage the switches separately. After the stack is set up, the switches in the stack
form a logical entity, and you can use a single IP address to manage and maintain the switches
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
7
iStack Technology White Paper
1 iStack
uniformly. The IP address and MAC address of the stack is the IP address and MAC address
of the master switch when the stack is set up for the first time.
Connection Topology
A stack has two connection topologies: chain topology and ring topology, as shown in Figure
1-9.
Figure 1-9 Stack connection topology
Master
Slave
iStack
iStack
Master
Standby
Slave
Slave
Standby
Chain topology
Slave
Ring topology
A stack usually uses a ring topology. In a ring topology, iStack blocks a stack link according
to the shortest path principle to prevent looping of data packets on stack links. A ring topology
is more reliable than a chain topology. In a chain topology, a link failure will cause the stack
to split. A ring topology will change into a chain topology when a link fails, so services in the
stack are not affected.
Adding a Member Switch to a Stack
During stack maintenance, the master switch collects topology information. When a new
switch joins a stack, the master switch is elected based on the following rules:

If the new switch has not joined any stack, the switch is elected as a slave switch without
changing the master and standby roles in the stack. For example, the new switch with the
iStack configured is power off and connected to a stack with stack cables. After the new
switch is powered on, it becomes a slave switch in the stack.

If the new switch has already joined a stack (for example, the switch has the iStack
function configured and connected to another stack using stack cables), the two stacks
merge into a new stack. The master switch of the new stack is elected from the master
switches of the two stacks. In the stack with the master switch elected as the new master
switch, all member switches retain their roles, and services are not affected. In the other
stack, all member switches join the new stack after restarting and synchronizing their
configurations with the configuration of the new master switch, and services are
interrupted.
Removing a Member Switch from a Stack
You can remove a member switch from a stack. The stack may be affected depending on the
role of the member switch that leaves the stack.
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
8
iStack Technology White Paper
1 iStack

If the master switch leaves the stack, the standby switch becomes the new master switch,
updates the stack topology, and specifies a new standby switch.

If the standby switch leaves the stack, the master switch updates the stack topology and
specifies a new standby switch.

If a slave switch leaves the stack, the master switch updates the stack topology.

If both the master and standby switches leave the stack, all slave switches restart and
form another stack.
Stack Split
As shown in Figure 1-10, a stack splits into multiple stacks when some member switches are
removed from the running stack with power on or when multiple nodes on the stack cable fail.
A stack may split into multiple stacks with the same configurations, which causes conflicts of
IP addresses and MAC addresses.
Figure 1-10 Stack split
iStack
iStack 1
=
iStack 2
+
iStack Link
Master
Standby
Master
Master
Dual-Active Detection
Dual-active detection (DAD) is a method to detect a dual-active scenario and take recovery
action, ensuring network stability.
DAD has two modes:

DAD in direct mode
As shown in Figure 1-11, DAD is performed between member switches in a stack using
a direct link.
Figure 1-11 DAD in direct mode
iStack
SwitchA
SwitchB
DAD Link
iStack Link
DAD packets

Issue 01
DAD in relay mode
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
9
iStack Technology White Paper
1 iStack
As shown in Figure 1-12, DAD is configured on the inter-chassis Eth-Trunk in a stack,
and DAD in relay mode is configured on the proxy device.
Figure 1-12 DAD in relay mode
SwitchC
Relay
Eth-Trunk
iStack
SwitchA
SwitchB
DAD Link
iStack Link
DAD packets
After a DAD link is configured, stacks exchange DAD packets on the DAD link. After a stack
splits into multiple stacks, the stacks compare information in received DAD packet with local
information. If the switch in a stack is elected as the master switch, the switch remains Active
and continues forwarding service packets. If the switch in a stack is elected as the standby
switch, the switch shuts down all its service interfaces except those excluded from shutdown,
enters the Recovery state, and stops forwarding service packets.
The rules for electing the master switch are as follows:
1.
The switch with higher stack priority is preferred.
2.
The switch with a smaller MAC address is preferred.
The election results are compared one by one. In the case of the same election result, the next
rule is used until the master switch is elected.
After the stack link recovers, the switch in Recovery state restarts and restores all the blocked
service interfaces.
Fast Stack Upgrade
Fast stack upgrade is a mechanism that upgrades the software versions of member switches in
a stack without interrupting service forwarding. This mechanism reduces the impact of device
upgrade on services.
During fast stack upgrade, the standby switch restarts using the new version, and the master
switch forwards data traffic. If the upgrade fails, the standby switch restarts and rolls back to
the previous version. After the standby switch is upgraded, it becomes the master switch and
forwards data traffic. The previous master switch restarts using the new version. After the
upgrade, the switch becomes the standby switch.
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
10
iStack Technology White Paper
1 iStack
1.4 Applications
Increasing Ports
As shown in Figure 1-13, when the port density of a stack is insufficient for increasing
number of servers, you can add new member switches to the stack to increase ports.
Figure 1-13 Increasing ports
iStack
iStack Link
Increasing Bandwidth
As shown in Figure 1-14, when the uplink bandwidth of a switch increases, you can enable
this switch to work with another switch to form a stack, and configure multiple physical links
of the two member switches into a link aggregation group to increase the uplink bandwidth of
the switch.
Figure 1-14 Increasing bandwidth
iStack
iStack Link
Simplifying Networking
As shown in Figure 1-15, multiple devices on the network form a stack and are virtualized
into a single logical device. The simplified networking does not require MSTP or VRRP,
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
11
iStack Technology White Paper
1 iStack
simplifying the network configuration. In addition, inter-chassis link aggregation implements
fast convergence and improves network reliability.
Figure 1-15 Simplifying networking
MSTP + VRRP
iStack
1.5 Example for Configuring the iStack Function
Networking Requirements
As the network size rapidly increases, the number of access interfaces provided by an access
switch needs to be increased, and the network must be easy to manage and maintain. However,
a single access switch cannot meet these requirements.
As shown in Figure 1-16, SwitchA and SwitchB form a stack, and service interfaces
10GE1/0/1 through 10GE1/0/4 are added to a stack interface.
SwitchA and SwitchB connect to SwitchC through Eth-Trunk10. Dual-active detection (DAD)
needs to be configured on Eth-Trunk10.
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
12
iStack Technology White Paper
1 iStack
Figure 1-16 Configuring a CSS
Network
SwitchC
10GE1/0/1
10GE1/0/2
Eth-Trunk10
10GE1/0/1~10GE1/0/4
iStack Domain 10
10GE1/0/5
10GE1/0/1~10GE1/0/4
10GE1/0/5
SwitchA
SwitchB
DAD Link
iStack Link
Common Link
Eth-Trunk
Procedure
Step 1 Configure stack domain ID of SwitchA to 10, and set the stack ID of SwitchB to 2.
<HUAWEI> system-view
[~HUAWEI] sysname SwitchA
[~HUAWEI] commit
[~SwitchA] stack
[~SwitchA-stack] stack domain 10
[~SwitchA-stack] commit
<HUAWEI> system-view
[~HUAWEI] sysname SwitchB
[~HUAWEI] commit
[~SwitchB] stack
[~SwitchB-stack] stack domain 10
[~SwitchB-stack] stack renumber 2
[~SwitchB-stack] commit
Step 2 Configure service interfaces 10GE1/0/1 to 10GE1/0/4 on SwitchA and SwitchB as physical
member interfaces and add them to a stack interface.
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
13
iStack Technology White Paper
1 iStack
[~SwitchA] stack
[~SwitchA-stack] port mode stack interface 10ge 1/0/1 to 1/0/4
[~SwitchA-stack] commit
[~SwitchA-stack] quit
[~SwitchA] interface stack-port 1/1
[~SwitchA-Stack-Port1/1] port member-group interface 10ge 1/0/1 to 1/0/4
[~SwitchA-Stack-Port1/1] commit
The configuration of SwitchB is the same as the configuration of SwitchA.
Step 3 Save the configurations of SwitchA and SwitchB, power off the two switches, connect the
stack link, and power on the two switches.
Step 4 After the stack is set up, check stack information. The following information shows that
SwitchA is the master switch of the stack.
<SwitchA> display stack
-----------------------------------------------------------MemberID Role
Mac
Priority
Device Type
-----------------------------------------------------------1
Master 0004-9f31-d520 100
CE6850-48T4Q-EI
2
StandBy 0004-9f62-1f40 100
CE6850-48T4Q-EI
------------------------------------------------------------
Step 5 Configure DAD in relay mode.
<SwitchA> system-view
[~SwitchA] interface eth-trunk 10
[~SwitchA-Eth-Trunk10] trunkport 10ge 1/0/5
[~SwitchA-Eth-Trunk10] trunkport 10ge 2/0/5
[~SwitchA-Eth-Trunk10] dual-active detect mode relay
[~SwitchA-Eth-Trunk10] commit
<HUAWEI> system-view
[~HUAWEI] sysname SwitchC
[~HUAWEI] commit
[~SwitchC] interface eth-trunk 10
[~SwitchC-Eth-Trunk10] trunkport 10ge 1/0/1
[~SwitchC-Eth-Trunk10] trunkport 10ge 1/0/2
[~SwitchC-Eth-Trunk10] dual-active proxy
[~SwitchC-Eth-Trunk10] commit
----End
Configuration Files

Configuration file of the stack
#
sysname SwitchA
#
interface 10GE1/0/1
port mode stack
stack-port 1/1
#
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
14
iStack Technology White Paper
1 iStack
interface 10GE1/0/2
port mode stack
stack-port 1/1
#
interface 10GE1/0/3
port mode stack
stack-port 1/1
#
interface 10GE1/0/4
port mode stack
stack-port 1/1
#
interface 10GE1/0/5
eth-trunk 10
#
interface 10GE2/0/1
port mode stack
stack-port 2/1
#
interface 10GE2/0/2
port mode stack
stack-port 2/1
#
interface 10GE2/0/3
port mode stack
stack-port 2/1
#
interface 10GE2/0/4
port mode stack
stack-port 2/1
#
interface 10GE2/0/5
eth-trunk 10
#
interface Eth-Trunk10
dual-active detect mode relay
#
return

Configuration file of SwitchC
#
sysname SwitchC
#
interface 10GE1/0/1
eth-trunk 10
#
interface 10GE1/0/2
eth-trunk 10
#
interface Eth-Trunk10
dual-active proxy
#
return
Issue 01
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
15
iStack Technology White Paper
A
A Terms, Acronyms, and Abbreviations
Terms,
Acronyms, and
Abbreviations
Full Name
Description
iStack
Intelligent Stack
Stack
Eth-Trunk
Issue 01
A A Terms, Acronyms, and Abbreviations
A technology of binding multiple physical
interfaces into a logical interface to increase
bandwidth, also called link aggregation.
Huawei Proprietary and Confidential
Copyright © Huawei Technologies Co., Ltd.
16
Download PDF