Switch Redundancy with 2848s

[jaredo_1]

Hello, I'm trying to do a redundant network setup (and learn more about procurves in the process). This is my proposed layout, utilizing procurve 2848s and probably cheaper switches for the access layer.

What would be the best way to set this up? MSTP?

We have a cisco asa with active/standby for a gateway that isn't included in this diagram, but would uplink to the 2 distribution switches. All servers would have redundant links to the server switches (lacp?).

Any inp9ut would be great.

[jaredo_1]

Since this will probably be a single vlan network, would rstp be a better route?

[Pieter 't Hart]

afaik MSTP can work together with RSTP.

MSTP means you can create multiple processes (instances) to determines the STP topology each for a different set of vlan's.
Using a single MSTP-instance for a single vlan offcourse is no problem

RSTP means rapid spannig tree. this modifies the behaviour of a port when a topology change occurs. Normally it first goes into a "listening" state when a change occurs (about 30 seconds). In listening state no packets are forwarded!

with RSTP the switch maintains a backup path to the STP root so it can immediately switch this backup path to forwarding state.
So traffic is resumed faster than with "old" STP (no wait/listening time).

LACP won't work accross different switches.

[jaredo_1]

So rstp will provide redundancy across the switches, but how can I do redundancy from the servers to 2 switches (with dual nics). Aside from running 2 separate vlan/networks.

[jaredo_1]

Basically I'd like the servers connected to 2 access switches, in which case one goes down they will automatically route through the other switch. These are all hp servers as well.

[Pieter 't Hart · Accepted solution]

The nic's of your server can be combined to form a "fault-tolerant team".
For HP proliant you use the "HP network configuration utility" to setup a team.
Other brands have similar tools.

(for HP) When you set the team to "auto" it can automatically detect it's connected to two different switches and configure itself into "fault tolerant only" (active/passive).
only one nic is active, the other takes over when the first link is down.
when both are connected to the same (801.2ag aware) switch it can form a "load balancing team" where both nic's are active.

[jaredo_1]

Perfect, that's what I though. :)

Thanks!

[jaredo_1]

Thank you!