Getting Started with vSphere 4 (iscsi, disaster recovery, storage)

[dforsman]

I have eight physical servers I would like virtualize into three physical servers. I have decided to use vSphere4 but have a few questions that the experts might know. We have 450Gb of active data. I also want to setup a DR site where I can transfer the virtual machines and data to a seperate server and power them up if required.

My questions are:

1. Does the storage of my data need to be on a SAN if I use the disaster recovery option?

2. I've also looked at the LeftHand and Falconstor virutal san products. Any comments?

3. When you configure the disaster recovery option is the data being transferred to the remore site get deduped so only the changes occur?

4. I heard that you cannot fail back a site? Is this true?

Any help or advice would be appreciated.

Thanks

[danpalacios]

The answers to your questions are largely "depends". I will try to give you some starting points, but the architecture will largely determine possilities.

1. Storage does NOT need to be on SAN. There are SANs which will replicate for you (important if you intend to use Site Recovery Manager). You could also use backup/DR products to pull the data from local storage (eg vRanger or PlateSpin)

2. I have not worked with either specifically, but they seemed to have interesting features. Again, there are many products which could fill your needs (Site Recovery Manager + Equallogic SANs, PlateSpin Forge + local storage, etc)

3. You would get this functionality out of a replicating SAN solution or a PlateSpin type option. A backup tool like vRanger would try to move the whole image every time.

4. Fail back is a function of your solution. With SAN replication, you can often reverse the flow of copying for failback. With the other options, you would need to reverse to copying based on the products built in tools. Site Recovery Manager supports failback among its scripts.

DR is complicated and largely driven by specific need and price. It looks like a mail server (you mentioned mx record) but details about budget, hardware, RTO/RPO, etc. If you have deeper/more specifc questions, I will help where I can,

[azn2kew]

As mentioned, what's your budget looks like and all the DR components can be automated using SRM and any storage vendor using there SRA for the replication. Even though, SRM doesn't have a button for failback but you can manually clean up and failback the process no problem. It would be nice to have a shared storage (iSCSI/NFS/FC) solution for this purpose, but you can still use local disk if you happy with the performance. You can create 3 host clusters and virtualize your expected machines and then use backup solution to have system state backup and find a means of transfering those backups to DR site. Standup several ESX hosts for DR and create VMX file for placeholder and restore the latest backup from DR when you wish. That's kind of manual process, it would be nice if you can afford SRM and its pretty easy and cheap with Dell Equallogic PS series which has great built in tools and can be configure in an hour. Again, its all depends how and what has been implemented but here is a great guide for BC/DR written by VMware team about 270+ pages. that should give you great information.

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

Regards,

Stefan Nguyen

VMware vExpert 2009

iGeek Systems Inc.

VMware, Citrix, Microsoft Consultant