Monday, June 3, 2013

Hardware considerations while using VMware Site Recovery Manager!

Today, I received a query from a VMware partner around hardware requirements for VMware Site Recovery Manager. Although the SRM solution is hardware agnostic in most of the cases, there are a number of things one must be careful of before implementing Site Recovery Manager in there vSphere infrastructure. I replied to the query and I thought I should share this with the community as well to ensure they can read this and make use of it.

Here is the Query

Can someone throw some light on SRM hardware requirements on both sites? Is there stringent requirements of matching CPUs at both end. While DR runs lesser capacity than DC, we want to plan lower performing CPUs. Can this different generation CPUs will be taken care by SRM??

Here was my Response


While your query holds good for traditional physical DR, with SRM the entire equation of hardware dependency changes drastically. Site Recovery Manager is a disaster recovery solution which is completely hardware agnostic as it operates at the software layer of the data-center which is vSphere.


To make it more simple, you can have any brand/model/make/CPU of server running in the Protected Site, and fail-over your virtual machines from this Protected site to Recovery site on a ESXi hosts which is of a completely different make, model or CPU from the Protected site. The only requirement is that both these server types should be supported with the vSphere version which you are running on them, hence they should be on our HCL (Hardware Compatibility List) for servers. 



Next, you need to look at the capacity of the DR site. As you mentioned, the DR site can be made up of less expensive hardware, however, at the same time, you should have appropriate capacity calculations to ensure that the Virtual Machines, if failed over to the DR site during a disaster or if powered on at the DR site in a Test Recovery Mode, should have sufficient Compute and Storage resources to operate normally and support the respective business functions.



Last and the most important dependency on hardware is for storage. As you are aware, SRM with version 5.0 and above supports both Storage Based Replication(SBR) and  Host Based Replication (HBR or vSphere Replication), you have a choice here. If you chose SBR, you need to have an identical storage array on both the sites and a storage replication software which will enable this replication. In this case, you need to check our  HCL (Hardware Compatibility List), to ensure that you a supported SRA (Storage Replication Adapter) available for that storage array. In case you go with the built in HBR or VR, you do not have to worry about the storage hardware again, since the replication happens from the VMkernel layer from source to destination host over the Networking Protocols. Also remember that you need to select the right version of SRM to go with the version of vSphere which you are running. Look at the Product Interoperability matrix for this.



To summarize, since SRM is an Active/Passive DR solution, it does not need you to run identical hardware and you can run any supported hardware for server configuration. For storage you have 2 options HBR and SBR. To read more about vSphere Replication and searching VMware Hardware Compatibility Lists, you can refer to the following articles I wrote a few days back:-








As always, Don't forget to Share and Spread the Knowledge. 


2 comments:

  1. Hi, I think you may have a typo when you are referring to the identical storage requirement. Your line says "If you chose HBR", shouldn't that be SBR (storage based replication) for identical storage?

    Mark

    ReplyDelete
  2. My bad... Updated that.. thanks Mark for pointing out :-)

    ReplyDelete