High Availability

chances of entire array going offline are very slim, it has to be something catastrophic like multiple drive failures, critical bug. For a solution that consists of 20 VMs i am not sure how much money you want to throw at it.  For a small environment like that i would look at VM based replication solutions like Zerto, EMC Recoverpoint for VM etc. If money were unlimited than you could look at  EMC VPLEX.

dynamo hit the nail on the head. For the size of environment you are talking about it is difficult to justify more than one array. We have a hard time justifying it in a much larger environment. if the applications are so critical that the business is willing to throw a lot of money at t a solution you might want to consider a smaller VMAX100k as an intermediate option between VNX and multiple arrays. And if they are willing to throw even more at it to make sure things NEVER go down then you would want a pair of VMAX arrays with a redundant VPLEX config in front of it.

Unless the application is beyond critical to the business that would likely be overkill. Personally I would first aim for something like the VMAX100K with 2 engines. That gives you many points of redundancy without going completely overboard. Of course you haven't mentioned anything about the performance or capacity requirements of the target solution either, and that may narrow things down significantly.

BTW - I'm not really bringing up Unity because I have no personal experience with that platform yet. if your capacity needs are smaller Unity might meet the needs as well as providing decent availability. I just can't vouch for it personally as I've never been hands on with it.

Hello Allen Ward

Thanks for replying

What do you think of 2 VNXe3200 and Native Replication ? We can perform Local and Remote Replication using that .

Are you familiar with this feature ? Is there any limitation or disadvantage ? 

Hello dynamox

Thanks for replying

I guess it'll cost a lot if we chose VPLEX . What do you think of 2 VNXe 3200 and native replication ( Asynchronous ) ?

Is it a good idea ? We can perform local and remote replication using native replication feature embedded in VNXe3200 .

oh yeah, VPLEX is not cheap and waaaaay overkill for what you are trying to do.

for a small environment like yours VNXe might be an attractive option (if budget allows and sized properly).  You must have read the paper but here is just in case.  Is async replication sufficient, that is for you and your business to determine how much data they willing to lose in case of a disaster (read about RPO).

https://www.emc.com/collateral/white-paper/h13685-vnxe-replication-wp.pdf

Thank you dynamox

I've already read some pages of this White Paper . I'm going to finish it as soon as possible .

Another question guys .

If I buy 2 Hosts and 2 VNXe , and install ESXi on Hosts . Should I present both Storage A and B to Hosts? I mean , If i present a LUN from SAN A to Hosts and it goes down, Should I manually present SAN B ( LUN B ) to Hosts to switch over? Or switching is automatically? BTW , Can you explain me fail over procedure ( when SAN A fails) in host side ? How Hosts start to read/write on SAN B - LUN B? Do storage boxes handle it or ... ?

when you say "SAN" do you actually mean storage array ?   SAN is a storage area network, it's your SAN switches, not your storage array.  There is no automatic failover between storage arrays.  If  VNXe 1 goes down for some reason, you will need to "fracture" replication (i don't know VNXe terminology), present VNXe 2 LUNs to your ESXi hosts, discover the datastore, import virtual machines and only then you can turn them back on.  Remember this will be crash consistent recovery, windows servers might need to run chkdsk, linux might need to run fsck, application like SQL might need additional recovery steps.  Basically your VMs going to be in the same state as if you yank power cords from your ESX servers.

I have never used SRM myself but i believe some of these steps above can be automated with SRM, take a look.

Configuring VMware vCenter Site Recovery Manager (SRM) with EMC® VNXe Storage

Thank you so much but SRM is used to handle NFS shared storage .

old paper

VMware Compatibility Guide - sra