Immediate Request Rapid PVST+, Etherchannel, HSRP

[csco11142276]

Experts,

I need a quick help from you guys. I am working on network redesign for my company where we are introducing 50 VoIP users and distribution layer. I have done a quick design where I am creating 2 stacks of 3750E's as distibution layer and 3750's for access layer for 50VoIP users. I have attached a topology for the distribution and access layer.

I have confusion over following points:

1). As in the topology shown that two GB Etherchannels between access and distribution, Do I still need to run Rapid PVST between access and distribution for load balancing or we don't need STP at all.

2). Do we still need to use HSRP at distribution layer stacks when access to distribution link is Etherchannel and doing load balancing of VLANs.

3). All servers are connected to core switches in resilient fashion, Do I need to run HSRP on core switches for Server VLANs.

Input to these ASAP would be a great help.

Thx a lot in advance.

[andrew.prince]

I my personal opinion:-

1) Running Rapid PVST would be a priority for me, Access to Distibution ideally is layer 2 only - so running rapid PVST is a must

2) I would not run HSRP on the distibution layer. Ideally you only need mgmnt IP addressing for the network devices here and the access layter.

3) I would run HSRP at the core layer - as this is where I would be running all my layer 3 processes.

Access - layer 2

Distribution - layer 2

Core - layer 3

Just my opinion.

HTH.

[csco11142276]

Thanks andrew. You are saying I dont need HSRP at distribution. The distribution switches are layer 3 EMI gonna run OSPF towards core. I was worried about gateway addresses and load balancing, so if I dont need HSRP at distribution then how gateway loadbalancing will occur for access layer switches.

[Kevin Dorrell]

Looking at a typical access VLAN, each distribution stack will have an SVI on that VLAN, so each will advertise an LSA for it into OSPF. If everything else is symetrical, the routers in the core should see equal path costs via each of the distribution switches, so load balancing should be automatic. I suppose it depends how your core switches relate to each of the distribution stacks, whether they would consider the paths to be equal.

I'm not sure that HSRP would help you there.

Kevin Dorrell

Luxembourg

[Kevin Dorrell]

Do I understand that the inter-VLAN routing will take place in the distribution stacks, and that the links between the distribution and access are layer-2. If so, my approach would be:

1. I would run rapid-PVST on each access VLAN. In that case, the channel between the distribution stacks is essential, and should carry the VLANs. At that point, you have a layer-2 loop on each VLAN consisting of the two distzribution stacks and the access switch(es). The rapid-PVST would sort out this loop. Make one of your distribution stacks the STP root, and the other is secondary root.

2. Yes, I would run HSRP on each VLAN between the two distribution switches (which are the routers. That way, hosts connected to the access switch will be able to find a router, even if one of the distribution stacks fails.

3. Your connections between core and distribution are working at layer-3, so HSRP is not needed on that connection.

4. If you have server VLANs implemented in the core (which I would not do ... I would give them distribution and access switches), then I would have HSRP running on those VLANs. But that depends also on your servers, and how they organise their redundant connections.

Kevin Dorrell

Luxembourg

[csco11142276]

Thanks Kev, What if I create server aggregation layer with similar set-up two Cisco 3750E's and create run HSRP there and advertise the learned routes to the core via OSPF. Would this make a better topology.

Also, in you opinion Etherchannel option wouldnt do the job properly between access and distribution.

Thanks