I am putting together an offsite disaster recovery plan for my organization and I have several options available to me. We currently use Veeam Backup and Replication for our in house backups. We are getting ready to stand up a secondary site in a separate geographic location with a 100 Meg MPLS circuit between the two sites. We also have identical blade clusters with vsphere and vcenter licenses for both, as well as redundant SAN environment. Everything is sitting at HQ, and we are thinking to ship out one blade cluster along with the second set of SAN (identical vendor\model\etc). I understand that SRM might be the best option for us since we can easily mirror our SAN across sites and tackle both HA and Disaster Recovery with the same solution...however I would like to avoid any further purchases at this time.
I would like to know what would happen, if I attached the mirrored datastores to the secondary vmware cluster. What would be the process to get those vm's up and running? Would they already be added to the inventory, or would I need to manually add them by browsing to the vmx files? Would this fail because the primary host cluster has claimed permissions for those vm's?
In the end, I don't need a fancy solution as we only have about 35 VM's, and since I have the ability to use array based replication I would like to take advantage of that without any additional software if possible. It may just turn out that my only option is SRM.
What do you think about native vSphere replication vs Veeam replication instead of mirroring the SAN? It just seems that is a less sophisticated solution than SAN - SAN replication if you already have the gear.
Thanks in advance for your input!
From my perspective, since you already use Veeam as your primary backup solution, you can start using it as a replication solution, as well.
In case of Veeam replication, it might be worth specifying proxy servers at both ends (primary, secondary location), so that, the traffic will cross the existing connection in highly deduplicated and compressed state.
Also, it stands to reason to deploy additional backup server at remote location, setting it responsible for replication jobs only. This way, if anything happens with your primary infrastructure (server goes down, for instance), you will be able to start required operations, such as Failover, Failback, from secondary location without any issues.
Additionally, in order to reduce the amount of traffic sent during initial replication cycle, you can put into use replica seeding functionality.
Thank you for your response. I agree that would be a very workable solution and I am very familiar with how to do that, however I am mostly interested in using the SAN-SAN replication in order to achieve the best RPO\RTO. Given that we have 100MBPS MPLS circuit which will be solely used for disaster recovery, we have more than enough bandwidth, so it is not a requirement to have highly compressed files sent across the WAN. If anyone can justify Veeam replication as the superior technology to array based replication (with or without SRM) I am open to hearing why, but ultimately I don't want to choose it just because its the lower cost option. We have identical hardware for both the SAN's and blades as well as an MPLS connection which is where most of the cost comes in, so I believe that we could have a very sophisticated solution, I just need to know if SRM is a requirement or not.
First of all, array-based replication is really great functionality, especially, if you can afford it (as we all know, it’s quite an expensive thing).
There are few points regarding SAN replication I’m a little anxious about. These include application consistency, failover/failback management and replicating corruption.
As far as I know, without usage of additional tools, SAN replication doesn’t provide you with proper application consistency. This might be a challenge, if you’re going to replicate VMs running special applications, such as Exchange, SQL, etc.
Also, unless you use Site Recovery Manager, the failover/failback orchestration will leave much to be desired.
Moreover, if you use SAN replication that provides only the latest infrastructure state, there might be some issues with data corruption having been already transferred to the DR site. In other words, if the disaster happens (be it data corruption, virus infection, etc.), with the “latest state only” SAN replication, you might face the situation when the corruption has been already transferred to the DR site and you’re left with “corrupted ” data on each site.
Anyway, as mentioned above, there is no denying the fact that the SAN based replication is amazing DR strategy.