2960 Clustering feature explanation.

[Michael Schwartzman]

Hi All,

Does anyone has any experience working with the Catalyst clustering feature?(Specifically on the 2960 series.)

I understood from the configurations guide that a single management IP address is being maintained for all of the cluster members.

I'm curios on the following:

    Configurations as vlan.dat, SNMP configurations, and global configurations would be synchronized across the different cluster members automatically when configured from the cluster master?

     To make switch/port specific configurations, will it be required to connect to each switch manually from the main "virtual IP address", or is there a centralized manner of configuring those using this feature.

Thanks in advance.

[daniel.dib]

You use stack cables to connect the devices in the backplane. This backplane can send more traffic than through regular ports.

When stacking you have a single point of management. So you connect to an IP and do all of the configuration from there. You can configure all of the ports from there. So if you have 3 switches the interfaces will have numbering like 1/0/1, 2/0/1, 3/0/1 and so on where the first number is the "module" meaning which switch number it is.

This is from the 2960-x configuration guide:

Switch Stack Configuration Files

The stack master has the saved and running configuration files for the switch stack. All stack members periodically receive synchronized copies of the configuration files from the stack master. If the stack master becomes unavailable, any stack member assuming the role of stack master has the latest configuration files.

The configuration files record these settings:

• System-level (global) configuration settings such as IP, STP, VLAN, and SNMP settings that apply to all stack members
• Stack member interface-specific configuration settings that are specific for each stack member

Note	The interface-specific settings of the stack master are saved if the stack master is replaced without saving the running configuration to the startup configuration.

A new, out-of-box switch joining a switch stack uses the system-level settings of that switch stack. If a switch is moved to a different switch stack, that switch loses its saved configuration file and uses the system-level configuration of the new switch stack.The interface-specific configuration of each stack member is associated with the stack member number. Stack members retain their numbers unless they are manually changed or they are already used by another member in the same switch stack.

• If an interface-specific configuration does not exist for that member number, the stack member uses its default interface-specific configuration.
• If an interface-specific configuration exists for that member number, the stack member uses the interface-specific configuration associated with that member number.

If you replace a failed member with an identical model, the replacement member automatically uses the same interface-specific configuration as the failed switch. You do not need to reconfigure the interface settings. The replacement switch (referred to as the provisioned switch) must have the same stack member number as the failed switch.

You back up and restore the stack configuration in the same way as you would for a standalone switch configuration.

Daniel Dib

CCIE #37149

Please rate helpful posts.

[Michael Schwartzman]

Hi Daniel,

Thanks for the prompt response.
However I was referring to a godforsaken feature called clustering not stacking.
http://www.cisco.com/en/US/docs/switches/lan/catalyst2960/software/release/12.2_55_se/configuration/guide/swclus.html

The motivation to try and use it is to save few bucks.

Thanks in advance :)

Sent from Cisco Technical Support iPhone App

[daniel.dib]

Wow, never heard of that before. Not sure how much would be gained from deploying that. Seems you still have to configure switches one and one although you can access them from the command switch.

What's the motivation for trying it? Easier management? From what I can read it seems the command switch acts like some kind of proxy and sends commands to members switches. Doesn't seem like the configuration is actually synchronized like with stacking. I can't be totally sure though. Here is a old Networkers presentation on it.

http://www.cnc.uom.gr/services/pdf/lan-switch/Switch_Clustering.pdf

Daniel Dib

CCIE #37149

Please rate helpful posts.