SAN - Prepare for the worst?

[CapnPugwash]

Hi,

 

Most of our servers are virtual and linked to a Dell Powervault MD3200i SAN. This now five years old & well out of warranty. It's a single point of failure.

 

Ideally I'd like a failover device so that in the event of disaster, the network didn't go down (some degradation of service would be acceptable) and I had a breathing space to sort the problem out.

 

I'm interested in peoples opinions / suggestions

[Gardinho]

I would go for another SAN. You tend to find that despite being a main datastore and strictly speaking, a single point of failure, they are designed to give you most redundancy.

 

You can configure 2 arrays on SANs with RAIDs so that VMs can easily be transferred from 1 array to the other in the event of a failure. The 2 arrays essentially have separate boards, batteries and network cards anyway.

 

That's my opinion anyway.

Edited October 6, 2015 by Gardinho

[3s-gtech]

Depending on loads, a decent NAS with iSCSI could stand in for a SAN as a cheap failover - obviously for much lower loads but could still do the job, without having such an expensive device as a SAN "just in case". Something like a Synology DS1515+ or similar, with decent processing power and plenty of GB ports.

[Gardinho]

Would it not make more sense to install a new SAN as the main datastore then use the old SAN as failover? They do have decent life as there isn't a lot that can go wrong.

[Asad_Symitry]

How many virtual servers are you running and what is the condition of the hosts that you have? How important is it to you as well to have no Single Point of Failure?

[rrrrr]

I looked into this when i was doing a server refresh. I didn't like the idea of having a single pint of failure. If the san backplane was to go or a dodgy firmware update was applied the whole network would be down. Failover san worked out to costly for the small chance it will be needed.

I decided to go for 3 servers with ssd storage and server replication between all 3. This way we can have a server go down and still function

[Asad_Symitry]

Do you have a separate ssd storage unit? Did you not look at a hyperconverged solution like Scale?

[rrrrr]

Do you have a separate ssd storage unit? Did you not look at a hyperconverged solution like Scale?

 

Yeah we did look at a solution like that. Cant remember the exact reason why we decided against it as it was a few years back.

Possibly as we was doing core switching at same time we got huge discount on hp bid pricing on whole solution and cost to spec worked out better with the solution we went for

[rrrrr]

Do you have a separate ssd storage unit? Did you not look at a hyperconverged solution like Scale?

 

No, we use ssd's in internal server storage, raided. The vm's are then striped across the 3 servers

[owenloughrey]

Have you had an opportunity to look at SimpliVity?

[Asad_Symitry]

Or Scale computing which seems to be Simplivity's cheaper cousin!

[mrbios]

EDIT: old thread