I’ve been disconnected from the internet the last couple of days because of a UPC screw up. So I missed out on all the SRM blogging. I am just reading all the new blogs that were created over the last couple of days. Here are just a couple random thoughts…
- The SRM docs talk about a “proteced” and a “recovery” site. Does this mean that SRM always needs an “active/passive” setup. In other words, can I only use 1 SAN for production.
Why? Well I could imagine that one would benefit from having some sort of load balancing with 2 VirtualDatacenters and 2 active SAN’s. Not only will the uptime increase, because if a site fails only half will failover, but also performance will be less of an issue having half of your active vm’s running on another SAN and cluster. And I’m not even gonna talk about TCO when half of your environment is doing nothing. - What about the VirtualCenter licensing? Do I need 2 VirtualCenter Licenses?
- What about failing back? I played around with a beta and there’s no automated option for a fail back at this moment, when will it be available?
- What about a third site? Think about switching datacenters every week or so for the insurance / banking industry. (Think Anti-Terrorism.)
Jon says
Yes , I think you cannot have a Active/Active SANs with SRM and you must have two VC.
(Datacore will do it for some money…)
You can see:
http://communities.vmware.com/docs/DOC-3210
BR,
Jon
Kyle Smith says
You can most definitely have active/active setups, assuming you’ve got adequate SAN storage to always be replicating the protected VMs.
As SRM deals with pairs of sites, you could probably have multiple backup sites, as long as you’re willing to create and maintain extra recovery plans (e.g. put one VM set in plan A that fails to site 1 and one set in plan B that fails to site 2).
I can’t speak to the other points you raise as my knowledge of the product is limited to what I’ve picked up from the demos and tech talks in the office. I should also point out that I’m speaking for myself and not VMware.
Rich says
SRM creates “shadow copy” VMs when you configure replication to the recovery site SAN. You are licensed from the primary site to the recovery site (shadow VMs) only. If you want to replicate to other recovery sites you need another license. If your recovery site is a branch office, for example, and you have VMs there that you want to replicate back to your primary site then that’s another license too.
Fail back after fail over is covered by your license, however. I asked specifically about that in the Partner Exchange Lab I attended last week. Unfortunately, the lab did not cover fail back so the technical details of how you do that are still fuzzy for me too.
More about SRM from my lab last week at http://vmetc.com/2008/05/08/vmware-site-recovery-manager-overview/
Toni Verbeiren says
As far as I remember from the questions and answers at VMworld Europe, these are some answers to your questions/remarks:
* The setup is typically active/passive, but nothing prevents you from running 2 SRM instances whereby one instance is active/passive and the other one is passive/active, such that both situations are covered. You pay twice a license for SRM.
* You DO need 2 VC licenses
* Automated recovery is not included right now, and also not easy. The response from one of the engineers of VMware was something along the line: SRM can not know what you have done between the failure and the time you want to restore. Maybe you do not want to restore the complete situation, but only a subset, for instance? That said, by going the passive/active ‘trick’, you can get around this.
* Speaking in terms of SANs, a third datacenter is not an easy thing to implement. I know of one financial institution that has an (EMC) SAN that is replicated accross three datacenters. It appears to have been the first case for EMC (at least in Europe).
Hope this helps!
Ernst says
Ok so how about this:
If you split a customer enviroment 50/50 over 2 DC’s with 2 VC’s and SRM running. Would it be possible to not only fail over site A –> B in case A fails but also B –> A if B fails? with the same SRM config.
This would be almost the same config as if you do a maual failback after a disaster.
Anyone idea’s?