Archive for January 4th, 2017

Boot from SAN / physical server RecoverPoint replica to VM

Wednesday, January 4th, 2017

2017-01-04 Initial Post

A few months ago I looked into this and could not find much information about it. In theory it seemed like it would work. I even asked two different consultants about it and neither had any experience with this. After testing multiple times, I can confirm that this works.

Goal:

Use existing EMC RecoverPoint (hardware appliance) to replicate entire physical server (OS and data drives) to DR site and then mount the replicated LUNs as RDMs to a VM at the DR site.

Currently the physical server uses a local direct-attached disk for its OS partition. It does utilize SAN LUNs for data so it already has connectivity to the SAN. One of the major tasks is to move the OS partition from the local disk to the SAN and then enable boot from SAN.

Result:

This works pretty much exactly as I expected. The VM boots up into Windows fine. DNS updates the AD DC/DNS server in the DR site and clients are able to access the VM. Failback also works, so any changes made while in “VM mode” will be seen by the physical server.

Why would you want to do this?

If you only have a handful of physical servers and you already use RecoverPoint and SRM to replicate VMs, there’s little justification for bringing in something like Double-Take or PlateSpin since those costs thousands and add more complexity and steps to your DR plan. Yes, you have to manually attach the VMs to the RDMs with my method, but you can prep some of this ahead of time with a placeholder VM so during actual failover it really only takes a few minutes per VM. And you could attempt to automate some of this to make it even quicker.

Hardware and software used during test:

  • Server: HP Blade BL460c Gen8 with QLogic QMH2572 HBA and Dell PowerEdge R620 with Emulex LPE1150 HBA
  • OS: Windows Server 2012 R2
  • Storage: EMC VNX5600 and VNX5200
  • Replication: RecoverPoint 4.1.2.3
  • VMware vSphere/ESXi: 5.5U2
  • Imaging: Macrium Reflect 6.1.1366

High-level steps:

  • Configure proper FC zoning, VNX LUNs, and RP CGs in both production and DR sites.
  • Install and use Macrium to image the local C: drive to a new SAN-based LUN C: drive.
  • Reboot the server and configure the HBA to boot from SAN (this is one of the trickier parts because each HBA vendor does it a different way).

To test, failover the RP CG then attach the replicated LUNs to the DR VM.

I don’t have time to detail ever little step, but any competent storage/server/VM admin will be able to figure them out. The point of this post was to make it known that this does work and is a viable option for DR.