Best practices for disk replication for DR site?

[The_Chadster]

We're in the process of building out an EV DR site in an offsite datacenter.  We have 5 mailbox archiving servers, 1 public folder archiving server, 1 journal archiving server, and 1 FSA archiving server.

My plan is to build out a VM-based DR solution using building blocks/USL.  The only phsyical server @ the DR site will be a standby SQL server.  The load that gets put on our SQL server I don't feel comfortable putting on a VM.  Unless there are major issues with our production datacenter, I'd expect this to be mostly read-only. 

My biggest questions are related to the disk replication.  Our disk is located on a Clariion/Celerra and we'll be replicating all this data to a similar unit in our offsite location.  Each of our archiving servers have 5 mount points of high speed RAID 10 disk for the index locations.  SQL databases are on the high speed disk as well, and the .DVS files get dropped on a Celerra NAS (under the same Clariion head).

I know when backups run that consistency is important, thus the requirement to put  EV in backup mode.  What is the best practice for replicating the EV indexes and databases and keeping everything in sync on the other end?  Obviously when backup mode is enabled, it won't be a problem, but if the Clariion were to catch fire during an archive run, how can I prevent my indexes, databases, and files on the NAS from being out of sync?  My SAN team is proposing a rolling snapshot/replication method where a consistency group is set for each server and they replicate one after the other and then start back over at the beginning.  They do not feel comfortable snapping and replicating as many LUNs as we have all at the same time for fear of what kind of load that may put on our SAN.  I am worried about this causing inconsistencies in case of a failure.

Has anyone else dealt with this or have any suggestions?

Thanks!

Chadster

[Maverik]

In all theory what you are trying is possible. However you need to be aware that the only supported tried and tested way is VCS or ms clustering. Ms clustering though I would have concerns with over a WAN. Based on the above it will be hard to be certain of any best practises especially if none are written by Symc.

[The_Chadster]

Thanks Don.  Does this mean the only Symantec supported method of doing this is with VCS or MS clustering?  Is there anyone out there who replicates their disk using mirrorView or otherwise instead of using clustering?

[Maverik]

Hi,

What this means is this is the tested solutions not that others will not work.  Its a difficult place to be in expecially if you did ever have an issue. 

As I said what you are trying to do sounds very very possible and I could not forsee an issue with it., as the replication etc if I undertand correctly will be transparent to EV. Therefore cannot forsee any reason SYmc would have a problem with it.  All I was mentioning is what is in Symantec Guides on what has been tried and tested through QA. Perhaps you could consider engaging Symantec Consultancy to review this plan and try and an offical reply from Symantec if they will rubber stamp this.

You sound like a fairly large organization, heavily involved with Symantec products, have you ever considered looking at Symantec Business Critical Services? The reason being is you be assigned a Business Crtical Account Manager and Busincess Critical Engineer to assist with projects like this.  Also you get a higher level of support. Just a thought.

[MichelZ]

Chadster

You could have a look at our product, EV NearSync. (http://www.evtools.net/products/ev-nearsync)
It's a "Universal" replication solution which integrates into Enterprise Vault, in the means of removing the Safety Copy (if enabled) only when the object was successfully replicated to the DR site.

Cheers
Michel