HSRP Vs redundant Sup

[NAVIN PARWAL]

Folks,

We have 6500's at the edge, distribution and at the core. We only have a single sup in the switches. We are doing HSRP at the distribution and the core. My question is that are there in good documents on CCO that talk about the advantages of redundant sup over HSRP? How can i convince my mananagement that they should buy redundant Sup rather than depending on HSRP.

[Prashanth Krishnappa]

I am not aware of any documents but some advantages of having redundant Supervisors that I can think off

1)Stateful failover. Hybrid supports High availability and Native supports SSO.

2)Easier to manage

3)More cost effective

But main advantage of redundant chassis is that it is useful in the unlikely event of an entire chassis going bad.

So, if I had my budget and the uptime was very important to my company, I would go with redundant cores with redundant Supervisors in each of them.

[mheusinger · Accepted solution]

Hello,

basically at the edge you loose the service completely (or partially, if there are backup connections in place) if the switch goes down.

In the end you have to decide, what is cheaper, a redundant sup or a violation of your SLA and angry customers potentially willing to change to another SP. This is no technical decision, but a financial or political one.

In general HSRP and redundant sups are all backup solutions. To justify the cost for redundant sups you would have to calculate the downtime (HSRP timers and MAC learning, etc. vs. switchover time) and compare the loss arising from downtime. Also take into account application needs. F.e. voip calls will be terminated by the users if there is a couple of seconds silence.

This might or might not be an argument for redundant sups.

In my opinion only a business case can give a good argument to justify the redundancy options.

To give an analogy: does everyone have a second car just in case the first breaks down? No, but there are backups, which usually are sufficient, like taking the bus. In case you are a taxi driver, things look different and you might need a "fast replacement contract" with your car dealer. In case you are Michael Schumacher and driving a Ferrari formula one car, the loss through an image damage for Ferrari justifies a second car (and an expensive one!).

Hope this helps! Please rate all posts.

Regards, Martin

[NAVIN PARWAL]

Martin,

Thanks for the awesome input, would you recommend Catos or Native? from uptime perspective and ease of management.

Thanks

[mheusinger]

Hello,

CatOS or IOS? Hm. I liked CatOS more when it comes to switching, but IOS is more common and for sure will be there in future. I am not sure about CatOS. It might really vanish and everything moved to IOS. It would align better with the general Cisco strategy of having a common CLI across all platforms.

Have a look at the PIX withvrsion 7 ...

Personal guts feeling: IOS.

Hope this helps!

Regards, Martin

[andrew.butterworth]

It is generally best practise to NOT have redundant Supervisors in the Distribution or Core layer switches. It is however best practise to have dual Supervisors in the Access layer.

This is due to what services you are connecting at each layer. If you follow the best practise design guidelines for a Campus network your Distribution & Core layers are fully redundant - i.e. if one of your Distribution switches fails its 'partner' will continue to route the traffic and end users will be unaffected (or at most minimally affected). Having dual supervisors in the Core & Distribution layers can introduce delays when failing over due to Layer-3 protocols converging, there is also a cost involved.

In the access layer you are connecting end devices and a switch failure will therefore directly affect users. Having redundant supervisors here reduces this risk slightly.

HTH

Andy