10/31/2016

Virtual Chassis Stacking

Table of Contents

» VCStack Introduction
» Connecting switches into a stack
» VCStack Configuration
» Software and configuration file synchronization
» Rolling Reboot
» Resiliency Link
» Managing Stack Members
» Monitoring and troubleshooting
slide 2

1
 10/31/2016

Click to edit Presentation title

VCStack Introduction

Definition of a Stack

► A switch made of several units

► A single IP address to manage the whole stack
► High speed stacking link
► All functionalities can be configured across the stack
► It is what we call Virtual Chassis Stacking

slide 4

2
 10/31/2016

Stacking Benefits
► Simplified Management
– A ‘virtual’ chassis can be configured /managed via a single IP address
• Reduces network administration overhead
► Simplified Configuration
– Often redundancy protocols like VRRP & STP are not needed
• Reduces management traffic on the network, and reduces configuration
complexity
► High Availability
– Network resources are spread across a number of stacked switches
• Reduced impact of any one switch failing
– In the core, access to resources guaranteed through resilient connections
• Available upstream to servers, and downstream to distribution switches

slide 5

Stacking Benefits
► L3 core stack
– Hardware redundancy
– High-bandwidth
– QoS
– Stacking
• High-availability
• Simplified management
• Resilient links
• Scalability
• Future proof
► L2+ Distribution switches
► L2 edge switches

slide 6

3
 10/31/2016

Stacking Benefits
► Resiliency
– Aggregated links configured across different switches in the stack
• Full bandwidth from all links available for maximum throughput
• In the event of failure, a connection to the network core is maintained
• High Speed Stack Recovery
– In case of stack failure, recovery typically occurs in less than 3 second
► Scalability
– Individual switches can be inserted and removed from the stack
► Low entry cost
– Pay as you grow
slide 7

Stacking Benefits
► Compatibility
– Complete stacking solution from one vendor
• Guarantees straightforward installation
• No interoperability issues with network infrastructure
• Service and support is simplified
► Future Proof
– A network solution can ‘grow’ as required
– A number of switches can create a powerful virtual chassis
SBx908 SBx8100 VCStackPLUS
X900 / X930 X510 /X510L
X600 / X610 X310
slide 8

4
 10/31/2016

Click to edit Presentation title

Creating a Stack

Connecting Switches into a Stack

► The proprietary high-speed communication protocol that is used over the stacking links
requires multiple twisted pairs and a high level of shielding.
► Specialized cables and connections are required.
► The types of cables and connections available are dependant on the type of x-Series
switches you are stacking :
– Back-port stacking on SwitchBlade x908
– Front-port stacking using 10G ports on CFC960 – SBx8100
– Front-port stacking using XEM-STKs on x900 series
– Front-port stacking using 10G StackXS or Back 40G modules on x930 series
– AT-StackXG slide-in modules on x600 & x610 Series
– Front-port 40G stacking on the DC2552XS/L3 series
– Front-port 10G stacking on the x510 series
– Front-port stacking on the x310 series
– Front-port stacking on the GS900 series

slide 10

5
 10/31/2016

Stacking on a SBx908
» On the rear of the SwitchBlade x908 chassis, there is a pair of fixed stacking ports.
» Back port stacking requires a specific cable (AT-HS-STK-CBL1.0)
» You have to order the cable separately of the chassis
» Two SBx908 can be stacked together
» Note that the cables are crossed over–port 1 of the top switch is connected to port
2 of the bottom switch, and vice versa

slide 11

Stacking on a SBx908
» This provides 80Gbps of stacking bandwidth per stacking port
» Total bandwidth between units 160Gbps
» Perfect for the enterprise core with very high-bandwidth

slide 12

6
 10/31/2016

Stacking on a x900
» Each XEM-STK module has 2 x 15Gbps HD stacking
connectors
» So, total bandwidth between units is 60Gbps
» You can stack up to two 2 switches
» You can stack x900-24XS and x900-24XT together
» You can stack two x900-12XT/S together
» You can’t mix x900-12XT/S and x900-24X in
a stack

slide 13

Stacking on a x600 / x610

► AT-StackXG slide-in modules on x600
► You can stack up to 4 x600 switches (8 x610s)
► You can mix any x600 in a stack (POE and non POE)
► Each stacking port provide 12 Gbps HD
► Total bandwith of the stack is 48 Gbps

slide 14

7
 10/31/2016

Stacking on a x610

► Up to 8 unit can be stacked with 5.4.3 firmware

► AT-StackXG or AT-x6EM/XS2 slide-in modules on x610
► AT-StackXG, has two full-duplex, 12 Gbps stacking
► AT-x6EM/XS2, has two 10GbE SFP+ stacking slots
► The two ports on an SFP+ AT-Stack module, AT-x6EM/XS2, are configured as
stacking ports by default. However, they can be reconfigured as 10 GbE network
switch ports.

LDS

Using the AT-x6EM/XS2 stacking module you can extend the distance between
stacked units to the maximum distance supported by the particular SFP you are
using. This capability enables you to create a stack of up to 8 geographically
separated x610 switches as a single stack.

8
 10/31/2016

Stacking on a x510
AT-StackXS/1.0
stacking cable
(includes 2 stacking modules)
UP TO 1 METER

slide 17

Click to edit Presentation title

VCStack Configuration

9
 10/31/2016

How the Stack Communicates

► The stack management uses a specific VLAN ID and an IP subnet, default
values are :
– VLAN 4094
– Subnet [Link]/28
► You may need to change these values if they clash with a VLAN ID or
subnet that is already in use in the network.
stack mana geme nt s ubnet < ip- addr ess>
stack mana geme nt v lan <2 - 409 4>
► The management traffic is queued to egress queue 7 on the stack link (no
other traffic should be marked for queue 7)

slide 19

Stack Roles
► Each switch in a stack acts in one role
– backup member (also called stack member)
– stack master (normally as the active master)
► The stack members are controlled by the stack master.
► The stack master performs a number of tasks that a stack member
does not perform:
– It controls all switch management activity
– It synchronizes boot release and configuration files with stack members
– All routing protocol packets are processed by the stack master. The stack
master then transfers any requisite table updates to the stack members.

slide 20

10
 10/31/2016

Stack Master Selection

► Master selection is based on two parameters
– Firstly - stack member’s priority setting
– Secondly - MAC address
► The switch with the lowest priority become Master
– Priority default is 128 - can change to select specific master
stack <switch stack ID> priority <0- 255>
► If several switches have the same priority, the one with the lowest MAC@
become Master
► Master selection is not related with unit ID ((ie master does not have to be
1)
► Any switches in a stack can potentially be Stack Master

slide 21

Virtual MAC Address

The MAC address assigned to a stack must be unique within its network.
► The virtual MAC address can be manually configured by specifying a VCStack
virtual Chassis ID of your chosen value - entered as a decimal number within
the range 0 to 4095. The value 120 is used in the following example:
a w p l u s (c o n f ig ) # s t a c k v i r t ua l -c h a s s i s- i d 12 0

The virtual chassis ID entered will form the last 12 bits of a pre selected
MAC prefix component; that is, 0000.cd370xxx.
a wp lu s ( co nf ig )# s ta ck v ir tu al -m ac

If you set the stack virtual-mac command without entering a value for the stack
virtual-chassis-ID, the switch will randomly select a virtual-chassis-ID from the
allocated range 0000.cd37.0000 to 0000.cd37.0FFF.
slide 22

11
 10/31/2016

Stack Member ID
► Each switch in a stack has an ID number, which can be an integer
between 1 and 8. The default on each switch is a stack ID of 1.
► The stack IDs on each switch within a stack are unique.
► The system can automatically assign a unique ID number to each
stack member
► Each member’s configuration is associated with its ID
– Allows putting the stack in a pre-defined configuration
– In case of conflict, system automatically modifies ID of the unit with the higher
MAC@

slide 23

Assigning Stack IDs

► Manual assignment on a switch before stacking
– s t a c k ( c o n f i g)# stack 1 renumber <1 -8>
► Automatic assignment as switches join the stack
– The stack master will be assigned stack ID 1, and the other switches will be
automatically assigned other IDs.
► Manual renumbering of a switch after stacking
– s t a c k ( c o n f i g)# stack 2 r enumber <1 -8>

slide 24

12
 10/31/2016

Stack Maintenance
► Adding a stack member
– A switch can be added to an existing stack (hot-swapped in)
• Power down the new switch
• Connect its ports to the stack
• Power on
► Removing a stack member
– A member can be removed from a stack (hot-swapped out)
• Power down the member
• Disconnect its stacking ports
• Reconnect the remaining stack members

slide 25

Stack Maintenance
► Replacing a stack member
– You can seamlessly swap a switch into the stack to replace another
• Configure new switch with the same member ID as its replacement
► Optional auto-upgrade
– Auto-upgrade will copy the master's software release onto new member
• If a new member joins a stack and has a SW release that is different, the master’s
software release is copied to the new member
• Auto-upgrade works when the master and new-member releases are similar (for
example 5.4.2-0.1 and 5.4.2-0.2)
– Auto-upgrade is enabled by default
• Disable with (no) stack <1-8> software-auto-synchronization
• If disabled, a new member with different SW release cannot join stack

slide 26

13
 10/31/2016

Provisioning
► Provisioning provides the ability to pre-configure ports that are not yet present in a
switch or in a stack.
► Provisioning keeps a 'placeholder' for a XEM or switch which has been hot-swapped
out.
► Switch provisionning
a w p l u s (c o n f i g)#switch 2 provision x900 -24
► XEM provisionning
a w p l u s (c o n f i g)#switch 2 bay 2 provision xem -12

slide 27

Click to edit Presentation title

Resiliency Link

14
 10/31/2016

Resiliency Link
» In this network scenario when
the stack link breaks, the edge
switches will continue to use
their uplink ports. This is a
problem because the two
switches elect themselves as
master and a duplicate ip issue
occur.
» The optional Resiliency Link
feature provide a solution to
this problem.

slide 29

Resiliency Link

► With a Resiliency Link, the stack members all listen for periodic
(one per second) Health Check messages from the master

► As long as the active master sends Health Check messages, the

other stack members know that the active master is still active.

► The Resiliency Link can be between

– Eth0 port (only on SwitchBlade x908 or x900 series switches)
– A dedicated VLAN (resiliencylink VLAN) to which switch ports may
become members. This is available on the x510, x600, x930

slide 30

15
 10/31/2016

Resiliency Link

Stack Master

High-Speed
Resiliency Link
Chassis Stack Connectors

Stack Member

slide 31

Resiliency Link Configuration

► The out-of-band Ethernet port is configured as a resiliency port with the
command:
awplus(config)# stack resiliencylink eth0

► Note that even if you configure the eth0 port as a resiliency port, you can still
use it for out of-band management.
► A VLAN, and switch port are configured for resiliency link connection with
the commands:
a w p l u s ( c o n f i g)# stack r e s i l i encylink vlan1000
a w p l u s ( c o n f i g)# interface port1.0.1
a w p l u s ( c o n f i g- i f ) # switchport resiliencylink

► This VLAN is dedicated to the resiliency link function and must not be the
stack management VLAN or a customer data VLAN.

slide 32

16
 10/31/2016

Managing StacktitleMembers
Click to edit Presentation

File Synchronization

► A VCStack requires that the software version and the

configuration files on all stack members are the same.
► The following files are synchronised by the stack master:
– Software release auto-synchronisation
– Shared running configuration
– Shared startup configuration
– Scripts
– Note : licences are not synchronized. For optional features (e.g. IPv6) each
switch in the stack must have his own feature licence

slide 34

17
 10/31/2016

Rolling Reboot

» This command allows a stack to be rebooted in a rolling sequence so that

no more than one unit of the stack is in reboot at any given time.
» First, the stack master is rebooted causing the remaining stack members to
failover and elect a new master.
» As soon as the rebooted Active Master has reloaded, it becomes the Active
Master again.
» Immediately after the Active Master has reloaded and assumed its role
again, all of the other switches in the stack are rebooted at the same time.

slide 35

Rolling Reboot
awplus#reboot rolling
The stack master will reboot immediately and boot up with the
configuration
file settings. The remaining stack members will then reboot
once the master
has finished re-configuring.
Continue the rolling reboot of the stack? (y/n):y
awplus#22:11:07 awplus VCS[995]: Automatically rebooting stack
member-4 (MAC: [Link]) due to Rolling reboot
URGENT: broadcast message:
System going down IMMEDIATELY!
... Rebooting at user request ...

slide 36

18
 10/31/2016

Remote Command

► There are some limited actions that you can do on an individual

switch member:
– Executing commands via “remote-command”
r em ot e- co mm an d <s ta ck I D> < op ti on >
r em ot e- co mm an d 2 sh ow s ys te m en vi ro nm en t
r em ot e- co mm an d 2 sh ow p ro ce ss

– Reload a specific switch in a stack

r el oa d st ac k- me mb er 4

slide 37

Remote - Login

► This command is used only on the master in order to log onto

the CLI of another stack member. In most respects the result
of this similar to being logged into the stack master.
► Configuration commands are still applied to all stack members,
but show commands, and commands that access the file system
are executed locally.
► To log onto stack member 2, use the following command:
► awplus# remote-login 2
► To return to the command prompt on the master stack member, type exit.

19
 10/31/2016

Monitoring and troubleshooting

► You can monitor and troubleshoot VCStack with several tools

– LEDs on the switch or XEM
– By using the show stack and show stack detail commands
– Stack debug output
– Counters
– Event logging

slide 39

Show Stack

awplus#show stack
Virtual Chassis Stacking summary information

ID Pending ID MAC address Priority Role

1 - 0015.77ae.60cb 128 Active Master
2 - 0015.77ae.5fdc 128 Backup Member
3 - 0015.77c2.4d56 128 Backup Member

slide 40

20
 10/31/2016

VCStack Implementation scenario

Enterprise convergence:

Resilience in CORE
Aggregated Links to Servers
Aggregated Links to EDGE

VCStack Implementation scenario

Resilience in CORE
Aggregated Links to Servers
Aggregated Links to EDGE

21
 10/31/2016

Steps to Set up a VCStack

► 1. Prepare the switches

► 2. Install and power the stack master
► 3. Install and power the backup member
► 4. Confirm that the stack is operating
► 5. Check the stack
► 6. Check the stack numbering
► 7. Configure virtual MAC address
► 8. Configure the stack as one switch

1. Prepare the switches

► Before connecting any of the switches together:

Ensure that all the switches have the same feature
licences installed.
If some stack members have feature licences installed for features
that will not be used on the stack, and other switches do not have
those licences installed, then remove those unnecessary licences.
If you are using XEM stacking, you must install the XEMs into the
switches before you can enter any stacking commands. An x900
Series switch will reject any stacking commands until a XEM-STK
is installed.

22
 10/31/2016

2. Install and power the stack master

► Install and power up the master switch. It will detect that there
are no other members in the stack, so it will elect itself master.
► Then, after the switch has booted, the show stack command will
show a stack with only one member:

3. Install and power the backup member

► Install the next switch, connecting the stacking cable from that
switch to the master.
► Note: Make sure the stacking cables are crossed over between
the stack members. This means that stack port 1 on switch 1
should connect to stack port 2 on switch 2. If this is not done,
the stack links will not come up and the stack will not form.
Power up the switch—it will detect that there is already an
active master, and so will come up as a backup member. The
active master will assign it the first available stack ID.

23
 10/31/2016

4. Confirm that the stack is operating

► Check that the stack links have all come up successfully. This
can be done by:
► checking the LEDs on the switches or XEMs. The port LEDs for
all stack members should be green. Port LEDs that are off or
flashing amber indicate that the stack is not operating correctly.
The master or status LED will be green on the switch that is
the stack master.
► using the show stack detail command. This command
provides a snapshot of the stack status.

5. Check the stack

► After bootup, the show stack command will show that there
are 2 switches in the stack:

24
 10/31/2016

6. Check stack numbering

► If you are not satisfied with the stack IDs that the switches have
been automatically assigned, then this is the right moment to
remedy that.
► To change a stack ID, use the command:

awplus(config)# stack renumber

It is usually a good idea to give the stack ID 1 to the active master,

as it is quite natural to associate the lowest ID with the active
master switch.

7. Configure virtual MAC address

► To manually select a virtual MAC address you enable the stack

virtual MAC feature by using the commands:
► awplus# configure terminal
► awplus(config)# stack virtual-mac
► awplus(config)# exit
► awplus# copy runing-config startup-config
► awplus# reboot
► reboot system? (y/n): y

25
 10/31/2016

8. Configure the stack as one switch

You are now ready to configure the stack with port channels, VLANs, IP
addresses, and so on. For example, to create a port channel that
aggregates ports from two different stack members, you would
configure as follows:
awplus(config)# configure terminal
awplus(config)# interface port1.0.1, port2.0.1
awplus(config-if)# channel-group 1 mode active
awplus(config-if)# interface port1.0.23, port2.0.23
awplus(config-if)# channel-group 2 mode active
awplus(config-if)#do show etherchannel summary

Remote login

► To login from the stack master (stack member 1 in this case) to

stack member 2:

26
 10/31/2016

Thank You!

Americas Headquarters | 19800 North Creek Parkway | Suite 100 | Bothell | WA 98011 | USA | T: +1 800 424 4284 | F: +1 425 481 3895
Asia-Pacific Headquarters | 11 Tai Seng Link | Singapore | 534182 | T: +65 6383 3832 | F: +65 6383 3830
EMEA & CSA Operations | Incheonweg 7 | 1437 EK Rozenburg | The Netherlands | T: +31 20 7950020 | F: +31 20 7950021

© 2016 Allied Telesis Inc. All rights reserved. Information in this document is subject to change without notice. All company names, logos, and product designs that are trademarks or registered trademarks are the property of their respective owners.

27