[THIN] Re: DR project dumped on me

  • From: "Webster" <webster@xxxxxxxxxxxxxxx>
  • To: <thin@xxxxxxxxxxxxx>
  • Date: Tue, 8 Dec 2009 10:53:24 -0600

The implementation engineer still has not replied with what is being used
for replication and the replication interval.





From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On Behalf
Of Jeff Bolton
Subject: [THIN] Re: DR project dumped on me


Not usually, but it depends on how the replication was set up.  Essentially
the SAN takes a PIT snapshot of the protected LUN and then replicates the
delta's to the remote site.  The replication cycle is usually based on the
defined RPO, so it could be anywhere between real-time to days.



From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On Behalf
Of Joe Shonk
Subject: [THIN] Re: DR project dumped on me

If replication is done at the block level, is the VM quiesced before
replication begins?  How often does replication occur?




From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On Behalf
Of Jeff Bolton
Subject: [THIN] Re: DR project dumped on me


If they are using Site Recovery Manager for VMware, the replication to the
DR site will be very easy.  SRM basically assists in creating a complete DR
plan for virtual environments.  It works in conjunction with the SAN.
Basically the SAN is set up to replicate a particular LUN or set of LUNs,
SRM then allows you to re-IP the virtual machines on the fly (if necessary)
at the destination site.  The destination machines remain powered off, until
a DR or test DR event is triggered within SRM, it then severs the
replication and powers up the DR VMs.

Because replication is done at the block level for the protected LUN(s),
there is no need for any additional data replication between the two sites.

If they're not using SRM with VMware, it gets a little more complicated. 

Hope this helps. 



From: Webster <webster@xxxxxxxxxxxxxxx>
Subject: [THIN] DR project dumped on me

Two days before "go live" I have had a DR project dumped in my lap.  The
original consultant is "no longer available".  I was not involved in the
planning, project scope, design, recommendations or purchasing of anything
involved in this project.  Customer has a main site and is implementing a DR
site.  They are using ESX 3.5 on HP C class blades with a 30MB pipe between
the sites.

For what I am responsible for, the main site consists of VMs for: 

SQL 2005 for data store 
License server 
2 CSG3.1/WI4.6 in NLB 
2 32-bit PS4.5 servers (Office 2007 and Acrobat Pro) 
1 64-bit test XA5/FP2 on 2003 server 

They want me to help them finish testing the 64-bit XA server and if
approved, migrate all users to the 64-bit farm and rebuild the 32-bit
servers as 64-bit servers.  Then they want all of this duplicated in the DR

I am not familiar with ESX and what it uses to move VMs between host
servers.  All hardware is identical between the two sites.

They will be using some VMware/EMC utility to replicate VMs and data between
the two sites. 

My question, if ESX has a utility that replicates the VMs between the two
sites and the DR site VMs will NEVER be "on" when the main site is
functional, what else needs to be done.  If the VMs in the DR site will have
the exact same IP addresses, computer names and configurations as the VMs in
the main site, all that should need to be done in a DR scenario is to point
the users to the new public IP of the CSG at the DR site?  Correct?

Does the SQL data store need to have the Publisher/Subscriber replication
setup?  Wouldn't think so in this case. 

What other info do you need to help me wrap my head around this since I am
supposed to start on this Thursday (and I will be at Ctx HQ all next week)?



Other related posts: