Has anyone rolled out SRM 5.0?

tbgree00

Now that my certifications are complete I'm moving on to my big projects for the year. SRM is easily the biggest one in the list and I know surprisingly little about it considered the webinars I've watched. I've signed up for the free VMware online training course and am watching those slides. I'm curious if anyone has successfully rolled it into production and if there are any good approachable PDFs, books, sites, etc that I should read.

I have the Admin guide and the evaluation guide on my kindle so I'm going to try those out. Is the VMware press book worth 35 dollars? Trainsignal has a SRM 5.0 course so I may subscribe for a month to watch that.

dave330i

You can get by with the pdfs. VMware press book gives you a nice deep dive into SRM and how to setup some of the SANs for array based replication. There's a 3rd party vendor, Zerto, which looks very interesting.

QHalo

Mike Laverick's book is all you need. I had zero experience and followed this book nearly exclusively with exception of my SRA documents from NetApp. Works like a charm.

Edit: Let me clarify that I installed 5.1 SRM on my 5.0 environment. With the exception of having to install the 5.1 client, it works just fine.

Mike Laverick’s SRM 5.0 Book | VMware vSphere Blog - VMware Blogs

tbgree00

Awesome to know about the PDFs. We already have SAN based replication on EquaLogic so we're good on replication as far as I can tell. I'll look into Zerto though. I just printed out the Admin guide in booklet for the flight out and will likely print the eval guide for the flight back so I can at least read the free official docs.

Thanks for your reply Dave!

Edit: I put in the request for my supervisor to buy the book so I will likely have it after my vacation. I'm glad to know it's a sufficient primary source. Thanks Halo

dave330i

Laverick's book covers EquaLogic, so you're good there.

slinuxuzer

I've deployed it a couple times here are some of the key issues that always come up.

1. Expectation of data loss - Make sure the customer understands what the RPO is, most of the time you will be using Async replication between data rooms and there will be a data difference between the two sets.

2. IP changes - Is there a layer two link between the data rooms? and can the Vlans in your primary room be trunked down to your hosts in your secondary room? if not you will have make the customer aware that triggering a SRM fail-over will require the IP's to change, make sure to investigate any application requirements that can't tolerate this, I work in a manufacturing environment and alot of our automation equipment can't tolerate this, so we have had to write our recovery plan in such a way that a fail-over requires moving this Vlan in routing to the secondary room, this also poses the problem that if your physical servers are sharing this Vlan, basically it's best if you seperate physical servers so they don't share a Vlan.

3. Vm recovery - Be aware that SRM recovers one VM at a time per host, this means if your recovery room as four hosts you can recover 4 VMs at a time, the general idea here is, when you need to add capacity to your vmware environment, it is likely better to add more hosts vs. adding more memory to existing hosts.

4. Databases - Try to keep each Vcenter/SRM database in the same data room as its application, I personally like to run SQL on the Vcenter box, also use full version SQL, rather than express.

5. DR plan testing - be aware that when you perform a test recovery, Vmare brings each guest up on a private Vswitch, so every Vm can talk to each other, but not the outside world, it's a good idea to have a domain controller in Vmware so you can have authentication services inside this "bubble" also if you don't want your customers to have to use the Vmware client to reach each recovered machine for testing, you'll need to build a test Vlan that is locked down through networking to only a few "test" clients.