Redundancy/Resilient Design

[George-Sl]

I have a design question for distribution layer,

if we choose to use catalyst 9600 for our distribution layer

https://www.cisco.com/c/en/us/products/switches/catalyst-9600-series-switches/index.html

do we still need to do a redundancy, by stacking two big 9600? since it has too many redundant powers already and two supervisor in case of failure, we still need to stack it with another one? why?

so I know this is not new, and I have experience doing the same thing for 6500, I just wanted to make sure in terms of latest best design practices still necessary to have redundant switches(even the latest 9600) in distribution?

[balaji.bandi · Accepted solution]

I totally agree with you, what you understand make sense here.

Since you have too many redundancy in the same chasis, why do you need another one to be part of  Stackwise Virtual or L3 redundent mode.

here is my suggestion and ask questios your self for business continuty.

1. what if the location wise failure.

2. Whole location has power outage.

3. Kind of DR plan.

4. Softwre upgrades and maintenance.

5. etc, so on

If above not very important to Business, and business not  five9 uptime if any of the above situation can cause outage to network.

Then you do not need another distribution kit.

[Joseph W. Doherty · Accepted solution]

Even with redundant power supplies, supervisors (and multiple line cards), the chassis, itself, is still a single point of failure as also is (somewhat) the OS. A second device increases your redundancy level. If you're really paranoid, you use a different model and/or same model running a different OS version.

Whether you need such a high level of redundancy, especially at the distribution layer, depends on your service needs.

BTW, the second device could be "sized" like a "doughnut" spare tire, i.e. it doesn't have the capacity of your primary device, but can keep critical functions going until the primary device is repaired.