As an independent IT consultant and subcontractor, Paul Winkeler is in the field regularly, helping IT managers find new ways to make their data center run more efficiently. As the owner of Ohio-based PBnJ Solutions, Winkeler recently began using virtualization as part of a disaster recovery (DR) plan for one of his clients and spoke with SearchServerVirtualization about the solution.
Virtualization technology is very popular for server consolidation, but what else can it be used for?
The concept of using virtualization in a disaster recovery environment is very high on the list. In the past, it has been expensive to get one server to replicate to the other because those two servers had to be basically identical for complete replication. In order for them to be identical, you needed the same hardware in both the main and backup locations. With virtualization, those hardware costs are cut down significantly because of the ability to host several machines on one server.
Can you give us an example of how virtualization can be used for DR?
One of my customers is a large investment company that uses SQL Server 2000 to track client activity. They need this information to be available at all times. We are doing something different for this client using virtualization to replicate that data. We have Microsoft Virtual Servers; one at the main site and one at a backup site. The virtual servers have virtual disk drives and we are using a product called LifeKeeper from Steeleye Technologies, Inc. to replicate the virtual drives back and forth. Anytime there is a change, the product reports the information to both servers, so the data is always up-to-date.
What are the advantages of using virtualization in this scenario as opposed to going with a more traditional DR solution?
For starters, one advantage to using virtual servers instead of physical servers is in the testing. When we were building the test environment, we basically had two virtual servers running on the same hardware. So we built the whole test environment at the client's home office without having to worry about having high speed connection between the home office and the backup site. I could just do all the testing locally.
Now that the servers are up and running, another advantage to having the data virtualized is that I can now test more frequently than I can with physical servers. With physical servers, testing is often painful and something that we try to avoid. Using virtual servers, it is easy for us to test because we can shut down and turn on these virtual servers easily. We can simulate server failure just by shutting them down. We don't need remote hardware access to power down
Any other additional cost savings you have seen using virtualization for DR?
I already mentioned hardware, but there are additional savings in that area. At the customer's main site, we have a ratio of 2 or 3 virtual servers per physical machine. But at the mirror site, we boost that ratio higher because most of the time those servers are really not doing anything mission critical; they are just sitting there on standby. This conserves the amount of money we have to tie up in a backup site. Previously, before we starting using virtual servers, there was a lot of capital tied up in the backup site that wasn't really necessary.
There is also a savings when it comes to personnel because virtual environments can be managed remotely, so we need much less personnel working on the management of the machines and they can now be used for other tasks.
For more on DR in virtualization, check out this tip on disaster recovery strategies for virtual environments.