I have a query re, a network I'm currently working on. Until recently the network was configured with layer 2 switches at the access layer with HSRP running at the distribution layer (3, running OSPF) for the data and VLAN's.
Recently they have moved over to using layer 3 (OSPF) at the access layer, with new data and voice VLAN's. Each new access switch stack (multiple 3750's in place of a single 4006) is therefore configured with legacy and new VLAN's.
No HSRP is configured for the new data VLAN's. Inorder to clean up the rationalize the HSRP configuration of rhte new and old VLAN's is it possible or neccessary to configure HSRP for the new VLAN's at either access or distribution layer since OSPF would take care of providing for link redundancy in the event of switch or link failure ??
If you only have end hosts in the access-layer that are singly connected to one switch then HSRP doesn't give you anything.
And there is no point in configuring HSRP in the distribution layer because the uplinks are L3 routed links so it wouldn't work.
And you are correct in that an uplink failure from the access-layer to the distro layer will be taken care of by OSPF. Actually it should be a lot quicker because OSPF will be seeing 2 equal cost paths anyway so if 1 fails the switchover is almost instantaneous.
HSRP should be useless on the new data vlans if they are confined in a single L3 C3750 stack as I would expect on a routed access layer.
You should think of moving the users of old vlans to the new ones in the meanwhile you can keep the HSRP groups on the old vlans.
HSRP is a protocol used within a single IP subnet to emulate a virtual router VIP at both L3 and L2: for the new vlans that are terminated on the stacks there's no chance to configure HSRP on the distribution switches they are not members of these new subnets.
>> since OSPF would take care of providing for link redundancy in the event of switch or link failure ??
this is correct OSPF will provide load-balancing and redundancy towards the distribution switches.
You could revert the reasoning: the HSRP groups are a sign of old Vlans spanning on multiple switches and terminated on distribution switches.
Hope to help
HSRP provides a virtual gateway, i.e. you would not normally deploy it for router to router communication especially using a dynamic routing protocol such as OSPF.
Within a 3750 stack, the stack can maintain a gateway based on the stack master. If the stack master fails, the gateway IP will migrate to new stack master. So, on the face of it, HSRP would not be needed on the very edge. However, there are several points to note.
First, when a gateway address moves to a new stack master, the MAC can change. (How fast or how "permanent" can be configured, I believe.) This may or may not confuse some hosts, especially since a gratuitous ARP, I believe is issued. If HSRP is used, the virtual MAC would not change.
Second, although rarely needed, I find it convenient to be able to place another router in service and migrate to it as gateway using HSRP while clients are active.
Third, I'm unsure whether the gateway migration feature is faster when the stack master fails using a "native" IP address or with HSRP.