SAN Replication/DR Newbie Question

I guess the answers all depend on what you are going to use for your replication, i.e., are you using Exchange DAG, Exchange CCR and/or EqualLogic Group Replication.

Answers are with regards to using EqualLogic Replication:

Q: If one replicates LUNs from a primary to a secondary DC, how does one make use of them at the secondary site?

A: Recovery from EqualLogic replicas (as in the event of a disaster), you can host the volume from the array recovery group and later failback to the original group, in what we call (in EqualLogic lingo) as Recovery mode.  Or you can Clone the replications (depending on how much space you have on the DR side array).  There is a complete section in the EqualLogic Group Administrators guide on the basic procedures for this.  See: https://support.equallogic.com/support/download.aspx?id=1502 (support account login required)

You may (and it’s highly  recommended), employ the the EqualLogic ASM/ME (Auto Snapshot Manager / Microsoft Edition) for creating Application consistent smart replicas. See media.community.dell.com/.../tr1040-asm-exchangeprotection_iso.pdf

Q: Say we had an exchange server in the primary DC with two LUNs presented from a SAN.  These LUNs host the Exchange DB and Log files.  Replication is configured to replicate these LUNs to a secondary DC.  Do I build a second Exchange server and have these LUNs presented to the server, but have the server shutdown during normal operations, and brought up only in the event of a disaster at the Primary DC?

A: Typically, on the DR side you would have some sort of way to bring up the replicated data from the recovery group.  So you would you would need some sort of duplicate host setup.

Q: If this is the case, does it have the same name at the Exchange server in the Primary DC?  How does Exchange handle this?  

A: That depends on how you deploy your exchange see this link for a quick overview: www.exchange-genie.com/.../database-availability-group-dag-exchange-2010

Q: Do we need to use boot from SAN so that there is also an OS LUN that is replicated, therefore turning the server hardware into a "dumb head" for want of a better word?

A: That depends on which direction of replication you go.  Typically, the longest (time consuming) is configuring the OS.  Having a consistent boot volume (and the same exact hardware on both sides) could shorten this process.

Q: Also, is it possible to make use of the LUNs in the Secondary DC in a read-only fashion so that we could run backups from these LUNs and not incurr a read penalty in the Primary DC?

A: Yes, you can setup your backup software to make copies of the replicated volumes.

-joe

In addition to what Joe said, I'd like to comment that you might be blending in "replication" with DR in that sense that with one, comes the other.  That is not necessarily the case.  You have some of those functions via ASM/VE and ASM/VE, but it is not a full business continuity failover solution like SRM.  

Remember that with the location you are replicating it to, you may very well be within a seperate IP netblock and perhaps domain name spaces.  So in the example of the Exchange Server, One could promote the replicas at the remote site as volumes (for both the data volumes, as well as the VMFS volume that house the exchange server.  But then you may have a different name space, IP schema, etc.

So take a step back and consider what your real intentions are.  That should offer some clarity.  In my case, our requirements did not include and instant-cutover, remote site turn-up.  If that were the case, I would have incorporated SRM.  We simply needed to protect our data, and our systems that serve up the data in the event something bad happening at the primary site.

I made a 5 part series on replication with an EqualLogic SAN.  You may want to check it out.

itforme.wordpress.com/.../replication-with-an-equallogic-san-part-1