redundant architecture with HP 2530 switches as core network layer

[simor]

Hello,

We need to update the network/system architecture of one of our subsidiaries.
1 HP switch  (let's say S0) is already on site and we bought 2 more HP switches (S1 and S2).
Model of the switches : 2530-24G-PoE+ - J9773A

We originally planned to reproduce the configuration of our main site, but the difference is that we have Cisco switches
with different features, and apparently we can't do the same thing with the specific HP switches we bought.
See the attached network diagram to have a quick preview of what it looks like.

S0 is located on the ground floor, S1 and S2 are on the first floor.  We first thought that we could stack S1/S2 but apparently the HP stack supported on these switches is only useful for administrative operations, thus we don't really need it This doesnt seem like a "true" stack.

The thing is that these switches don't support a lot of L3 features, such as HSRP, which we have running on our main site onto the core network stack, and we can't have a real stack working (this is maybe 2 of the main issues).

As you can see we have 2 ESX servers, each with 2 links towards S1 and 2 links towards S2 (HP trunk, lacp).
The default gateway of the servers (VMs) we already have  configured is : 10.12.230.250
Vlan 230 = servers vlan, 10.12.230.250 = ip address of SW1 onto that vlan.

SW2 has another ip address on that vlan, but how could we use it as a gateway without a real stack or HSRP running in background ?

My question is : how could we configure the switches or change the architecture in order to have a redundant architecture, notably regarding the traffic distribution ?

Currently, even with the 2 1st floor switches, if SW1 goes down, traffic from the servers will not be routed.

Maybe we now have to look at the servers configuration ? Set 2 gateways on each server ? doesnt seem really optimal to me.  Would load-balancing be possible this way too ? ... I'm quite running out  of ideas.

Thanks if you have any input with this situation.

Regards,

M.

P.S. This thread has eben moved from Switches, Hubs, Modems (Legacy ITRC forum) to ProCurve / ProVision-Based. - Hp Forum moderator

[Vince-Whirlwind · Accepted solution]

Why use those switches as your servers' gateway?

Who not use Switch0?

Just use S1 & S2 as Layer2 switches only?

Having said that, I don't think the 2530 is a totally suitable choice for either core switching functions, Layer3 functions, or as a datacentre switch supporting ESX servers.

And no, the 2530 does not stack.
2920 does stack.

[simor]

Thank you for your response.

You're certainly right regarding the switch model not really matching our needs.

The ESX servers are on the 1st floor and will be plugged onto S1 and S2 (as indicated on the diagram).

The idea of having 2 switches doing the L3-routing was to have a redundant system, if 1 switch goes down, the other one could still route the packets. In our main site it's a 3-switches stack which is responsible for this routing, so we do not encounter this situation.

I guess we could use the firewall as the servers' gateway too.

Regards,

M.