We implemented in our DataCenters a solution with 2x3650 10G as redundant cores and 2960S as the access layer.
It includes Many Vlans, VRF to each costumers, and HSRP between the two 3560 to provide the High availability.
The solution works good, if the primary 3560 goes down the other one assumes with very low losses...
But, the problem is in the maintenance synchronism of the config of the both 3560.
- Things like ACLs, Routing protocol, standby IP Address, etc... NEEDS TO BE IDENTICAL.
- Things like Physical Interfaces configuration, IP address of Logical Interfaces, etc... NEEDS TO BE DIFFERENT.
We agreed to make changes preferentially on the primary 3560, and make a "show archive config diff" on the secondary to analyze the differences and manually aply it on the secondary.
It helps, but still is too much complex to maintain the environment.
One of the linuxers boys in here suggested to write a tool to compare the configs, discard the differences that needs to be different, and inform by e-mail those that need to be applied.
It maybe works, but to me it does not look good.
Does anybody have a better suggestion?
Message was edited by: Douglas Fernando Fischer I wrote 2560S, and I corrected to 2960S...
Unfortunately, the 3560 series is not stackable. In order to have identical configs, you would need a couple of 3750s stacked. With that, you only apply the config to the primary and the primary will sync the config to the rest of the switches.
I am not sure that there is way to do this by email.
We were aware of the solution based on 3750 and stacking when we decided by the 3560.
What I forgot to say is that those 3560 are in different buildings, connected through fiber.
So, in this case, the option that would solve the H.A of gateway and the syncing of configuration would be based on 6500 and VSS. And then the price would be very different.
Reconstructing our environment in a lab to develop a tool in linux which helps with syncing, I did something I never do (and I believe most of the guys do not do also), I entered in the "initial configuration dialog. "
I got to a point where he asked me about a "cluster command switch" .
And then came more questions:
- Is this is the tool that will help me in the config syncing problem?
- Has anyone used this for real or is this one of those features that nobody uses?
- It controls the routing part, too, or just part of LAN switching?
Switch clustering , will not help you syncronizing two or more switches.
The purpose of the switch clustering is "you can configure and troubleshoot a group of different Catalyst desktop switch platforms through a single IP addres".