SAN Mirroring

Originally Posted by Vanilla
What protocol? iSCSI, NFs, FC?

Dont forget the protocol type!
If iSCSI youl need to look at your networking kit and potentially need to beef up your switches to provide dedicated iSCSI ports/mezzanine cards/switches.
If FC then youl need to invest in fibre channel switches, HBAs in your hosts etc.

If your going to be hosting your production servers on these hosts and a SAN then FC should be the way you go, make sure your switches/HBAs are 8GB to leave you room for growth as in server sprawl and performance growth.

P.
 
BoomAM said:
From talking with suppliers, its looking like Equalogic or Lefthand at the moment.
With EMC being significantly more expensive...
Click to expand...

Dont go with Lefthand... not very good for VMware unless you opt for 10Gbe option.
 
BoomAM said:
From talking with suppliers, its looking like Equalogic or Lefthand at the moment.
With EMC being significantly more expensive...
Click to expand...

That's just not true any more at all. 12 months ago sure, VNX has been priced in line with the market now. Yes I am an EMC employee :) I haven't come up against NetApp, Dell, HP, HDS or IBM this year where we haven't been bang on with pricing without dropping our pants.
 
I'd be more inclined to buy one SAN with full redundancy built in.

Having 2 in the same location isn't really providing you much extra redundancy. For the first one to fail you'd more than likely need a power failure and if that happened the second one would be useless anyway.
 
Don't see the point if they're literally two metres apart - probably better to get a decent backup solution in (Veeam) and do it that way (you can do replication with that and it's quite cheap - just add another LUN on the other SAN into VMware and setup Veeam to do it that way).

Are you talking purely about the storage dying or are you talking about the VM's dying? If it's the VM's you could use Fault Tolerance on VCenter. Not perfect but beats buying a new SAN.



M.
 
Andy100 said:
I'd be more inclined to buy one SAN with full redundancy built in.

Having 2 in the same location isn't really providing you much extra redundancy. For the first one to fail you'd more than likely need a power failure and if that happened the second one would be useless anyway.
Click to expand...

Plenty to go wrong with a single SAN, backup unit is far preferable to resilience in my experience, we've had two failures this year where supposedly fully redundant, resilient anything kit managed to complete die or eat it's data. We'd have been stuffed without mirroring...
 
got to say, Equalogic has worked out very well for our esx setup. was a little apprehensive about iscsi support only and the ms initiator (exchange raw lun) but it really does work. would not put it with dell powerconnect switches though.... now where's my waffle iron?
 
bigredshark said:
Plenty to go wrong with a single SAN, backup unit is far preferable to resilience in my experience, we've had two failures this year where supposedly fully redundant, resilient anything kit managed to complete die or eat it's data. We'd have been stuffed without mirroring...
Click to expand...

Which SANs? With a HP EVA you'd have to be very unlucky to lose a disk group yet alone the full SAN.
 
MilanoChris said:
Which SANs? With a HP EVA you'd have to be very unlucky to lose a disk group yet alone the full SAN.
Click to expand...



Various units, they all suffer from firmware bugs to a greater or lesser extent, we have cases open for bugs with Nexsan, Hitachi and Netapp currently, there have been others in the past.

Hardware resiliency became a lot less important than code stability a while back when it comes to uptime. Same is true of most devices today - routers and firewalls are just as culpable, n+n resiliency isn't worth anything if the code is rubbish and as more and more features get rolled in there ends up being more and more to go wrong.

A Nexsan with dual controllers had a memory leak last week, crashed and triggered a fail-over, the secondary controller then managed to completely trash the host ID table. 40 Minutes to figure out and fix that, if it'd had been single controller it would have rebooted and come back up fine in less than 10% of that time....the irony...
 
New EMC VNX units with remote protection suite.

I've never had any problems with dual controller EMC units, but for DR purposes it's always nice to have a secondary unit.
 
You must log in or register to reply here.
Back
Top Bottom