cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1245
Views
30
Helpful
15
Replies

CST with HSRP

Patrick McHenry
Level 3
Level 3

Will I have a problem with CST spanning tree if I divide the vlans between two cores(making one the gateway for odd vlans and the other for even vlans).?Some of our old swicthes(3500 XLs) only support CST.

Thanks, Pat.

1 Accepted Solution

Accepted Solutions

No problem - not wearing anything out

Only reasons for me would be if you had different switch models at the core, perhaps if one switch was on newer hardware, or if one of the switches was significantly more reliable than another etc - I have seen this before where a company just did not have the capital to upgrade both core switches at the same time and they were worried that the older hardware was becoming unreliable - cards occasionally needed resetting.

Also if traffic utilisation across the links is minimal when all VLANs are going using one path and CPU/Memory utilisation is low there might not be a specific reason to load-balance - although I would still go ahead and use both core switches because as the demand on the network grows you then don't have to worry about re-configuring everything in the future.

View solution in original post

15 Replies 15

Dennis Mink
VIP Alumni
VIP Alumni

Pat,

when you say "gateways for odd vlans....." do you actually mean have one switch as the root for a set of vlans and the other switch for the remaining vlans? by means of altering the priority between the 2?

Please remember to rate useful posts, by clicking on the stars below.

yes, that is what I mean. At present, only one of our cores is the active for all VLANs and the other is the backup.

Thanks for responding.

Hi there

You won't have a problem as such but it will mean that you might have a situation where you have a less than optimal topology for some VLANs in your network which are configured on the 3500XL switches. It depends on how the switches in your network are connected together and where the 3500XL switches are positioned within the topology.

Hope this helps - do you have any diagrams? I am happy to take a look for you if you want.

Bryan

Thanks, Brian.

I'll try to draw something up. Alot of our data closets don't have 3500s. Hopefully we can get rid of the rest sometime soon.I think from what you are saying, it would probably be beneficial to have any 3500s in the middle of the stack. Is that correct? If they were the switch that actually connects to the core, the path might have to go through the whole stack of switches? I'll draw a diagram.

Pat.

If they are in the middle of a stack and the switches connecting into the core do support PVST (would recommend Rapid-PVST if possible) then you won't have a problem at all and it makes no difference that they only support CST. The only ports active on thise switches will only be connecting to end hosts anyway :-)

Bryan,

I attached a doc. It would be great if you could look at it. Let me know if it is clear enough. And yes, all other switches will be pvst or rpvst. Thanks, Pat.

Hi,

Looking at your diagram I think you would be better to have the 3500XL switch connectint directly to the core and have the 2 x 3560's with one connecting to the core and 1 in the middle. This way the switch in the middle should block the ports for the VLANs and provide the load-balancing you want n the same way that the switch in the middle is the one blocking ports on the left-hand side of the diagram.

Hope this helps.

Bryan

Going through different scenerios, it seems that it doesn't really matter where I put the 3500s. Would it?

Thanks for your help.

Hi,

No problem.

My mistake - I thought you only had one 3500XL. In that case no it won't - you will get the same result wherever you put them. I wouldn't worry too much about it, for 1/2 your VLAN's it just means that they will have to be switched to the second core-switch to reach the gateway, but its across a 20Gbps Etherchannel anyway so you won't notice anything. Just try and convince your company to replace the 3500XL switches if you can given that they are long past end-of-life anyway :-)

Thanks. Would you agree that this would be the better way to set up the network, rather than having one core do all the routing? Another thing, we want to cable each server switch in the data center to both 6509s. I think that is where we would get most benefit, having one link to one core fore odd vlans and another link to the other core for even vlans. As far as the 3500s it seems like every other month one of them dies anyway. They are leaving by attrition. Thanks again.

Yes, unless there are specific reasons not to then I prefer to try and load-balance traffic between 2 core switches and your network topology is typical for doing this.

Something else you might want to consider is reducing the HSRP times to speed up convergence if you have not already done so, the default is 3 seconds (hello) and 10 seconds (hold) and I change these to be 1 second and 3. It is possible to have shorter timers than this but I have found that you can have issues where the standby switch becomes active after not recieveing hello's within the hold time.

Thanks. I hope I'm not wearing out my welcome but, what would be reasons why you wouldn't want to load-balance? My boss is a little leary of the change.

No problem - not wearing anything out

Only reasons for me would be if you had different switch models at the core, perhaps if one switch was on newer hardware, or if one of the switches was significantly more reliable than another etc - I have seen this before where a company just did not have the capital to upgrade both core switches at the same time and they were worried that the older hardware was becoming unreliable - cards occasionally needed resetting.

Also if traffic utilisation across the links is minimal when all VLANs are going using one path and CPU/Memory utilisation is low there might not be a specific reason to load-balance - although I would still go ahead and use both core switches because as the demand on the network grows you then don't have to worry about re-configuring everything in the future.

Thanks for all the help!

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco