Active/Active Sites (vSphere FT and HP P4000)?

[hutchingsp]

http://www.youtube.com/watch?v=6n3VpfesM8A

I've been looking at SAN/storage options and have been thinking very much in terms of "Live" and "Backup" sites. Both our locations are around 3/4 of a mile apart linked by 1 or 10gig ethernet (the fibre will take 10gig so it comes down to cost and necessity).

I'm now thinking why on earth wouldn't I want to go active/active if we're looking at Lefthand? I'm hoping to have them in next week to discuss options.

I'm a little blown away by that video tbh as it's not using expensive license options.

Appreciate any thoughts on whether vSphere's FT really is good enough that you'd be content with (for example):

Primary Site:

ESX Host

P4000

Single Switch (presumably with VLAN's to keep iSCSI and vMotion/FT traffic separated?)

Secondary Site:

ESX Host

P4000

Switch (presumably with VLAN's to keep iSCSI and vMotion/FT traffic separated?)

Thanks (I have also posted this in the Enterprise forum but would be grateful if you'd let this one remain as IMO I'm aiming at people who may not look at the Enterprise forum but are actually doing this).

[Borja_Mari]

Hi,

i would recommend you to check this.

Hope it helps

Regards/Saludos,
Pablo

Please consider awarding

any helpful or corrrect answer. Thanks!! -

Por favor considera premiar

cualquier respuesta útil o correcta. ¡¡Muchas gracias!!

Virtually noob blog

[dragooner]

FT still has a lot of limitation.Most of the product environment could not use it,because of the performance.

[mcowger]

Not sure what you mean about performance - FT has a sub 10% hit for most applications. Granted, its not as useful right now as it could be given the no SMP limitation, but thats functions, not performance.

As far as going active active with FT cross datacenter - be aware there is more to FT than bandwidth - you need to be aware of latency as well....

--Matt

VCP, VCDX #52, Unix Geek, Storage Nerd

[hutchingsp]

Not sure performance should be too much of an issue for us as we're not CPU bound and right now all our VM's are single vCPU anyway.

I wouldn't be looking at doing this on all VM's but for core services (thinking Exchange in particular) this looks like it has potential vs. paying for a bunch of kit to sit at a backup site doing nothing other than waiting for the primary site to fail.

Any input on how the switching would normally be done in this kind of scenario?

Regarding latency, I'm not a networking expert but we'd be talking about a dedicated fiber core just for this stuff and the existing link is < 1ms just from doing continuous ping between a couple of servers over the existing fibre.

To put the VM side of things in a little context we're "only" talking about a couple of ESX hosts total, IOPS at 95th percentile is around 2000 (for everything we have).

[mcowger]

With plenty of BW and sub 1 ms RT times, you might get reasonable performance. Worth a shot!

--Matt

VCP, VCDX #52, Unix Geek, Storage Nerd

[dragooner]

In my opinion,FT does not support SMP is not a fouction problem.The single vCPU will cause performance problem Self-evident.