Backing up data is a lot more complex in VMware virtual machine environments. Jeff Boles, senior analyst and director, validation services at Hopkinton, Mass.-based Taneja Group, discusses the challenges of protecting data in virtual environments
Q: What are the challenges when protecting virtual machines?
A: Server virtualization has created a whole bunch of interesting challenges for the backup team. You have a whole bunch of physical machines that you've moved onto virtual machines [VMs], and while this might be good for making better use of your underutilized hardware, the one time your old hardware wasn't underutilized was when you were doing backup, and you were simultaneously hitting all these individual systems with high-speed backup streams. Now you have all these machines sitting on one piece of hardware as virtual machines, and you can slaughter that physical machine with the demand of all those backup streams.
Q: What are the best practices for virtual machine backup straight out of the box?
A: I have four solutions I consider out of the box. The first is snapshots over NAS [network-attached storage] or SANs [storage-area networks], which are pretty complex when it comes to virtual machines because you have this whole other layer of software you have to quiesce and integrate. Second, backup software on the VMs themselves. Third, there's this concept of service-console backup, which isn't very widely used anymore. It moves the backup software itself down to the service console and hypervisor level, but you're still gaining the same issues of I/O contention and limiting yourself to backup image recovery when trying to do restorations. Fourth is the proxy-backup method, which around VMware is called VCB [VMware Consolidated Backup] proxy and is considered state-of-the-art. VCB proxy introduces another virtual or physical machine into your environment, which acts as a proxy for accessing virtualization environment snapshots and backing them up with backup software.
Q: What do you look for in a third-party virtual machine backup solution?
A: Frankly, I expect VMware to continue to make big moves in this market, but there's going to be a ton of competition. There are some third parties out here really doing cool stuff. First, a solution should give you best-of-class backup at the virtual machine level. This means you can see what's inside a virtual machine, back up anything you want directly and restore it directly. Second, you should have the best-of-class capabilities outside the virtual machine as well. That means including things like creating synthetic fulls from differential data backups. Third, you should be able to do all of this while reducing and optimizing your I/O demand at the client. If you can reduce your I/O, you should be able to do things like concurrent backup. Fourth, you should still expect integration with your overall enterprise backup approach.
This was first published in May 2009