Redundancy strategy for 2848's

[esmythe_1]

We have two 2848's performing core and access switching, and inter-vlan routing for our san and cpu nodes.

Since the 2848's do not support xrrp/vrrp, what other strategies are there to get some redundancy?

Is it practical to use two nics on each server with separate vlan's to separate switches and trunk the switches?

[cenk sasmaztin]

2848 series switch unable L3 redundancy
therefore you must be only L2 redundancy with
spanning tree

or

connect each server,and each edge switch with two interface either 2848 switch

redundancy desing must be your network topology

please talk about more info your network

[esmythe_1]

Hi Cenk,

We have two storage nodes in a SAN and two cpu nodes for our virtualized infrastructure. We currently do not have enough devices to warrant separate core/edge switches.

I would like to configure the 2848's in the best possible way for failover. Meaning, if one switch dies, the cpu nodes can continue to access the SAN.

I had intended the SAN to be on VLAN1 and the serves on VLAN2, but was using the core switch for routing. Since the 2848's are 'lite l3', I can't use a vrrp or hsrp-equivalent protocol to make sure the gateway is accessible in event of switch failure.

My question is if there was another way to design the network so that there would be failover in case of switch failure.

thx,
-e

[Pieter 't Hart]

if you really need redundancy for your router, it's probably the simpelest to add devices (like two dedicated routers) that do support a redundancy protocol.