Server virtualization delivers clear benefits, including a greater return on your server investment and a reduction of power and cooling costs. But it also presents a number of challenges. In particular, virtualization complicates backup and disaster recovery. While the traditional methods for backup will work, you now have much better options.
Whether virtualized or not, backup and recovery is about protecting your data. Data protection generally has four stages.
Stage 1—Using SANs makes it possible to backup storage shared by multiple servers and applications in a single, centralized backup operation.
Stage 2— Adding extra disk capacity to expand SAN backup improves both Recovery Point Objectives (RPO) and Recovery Time Objectives (RTO) through the use of mirrored disk break-offs, or a 2-stage backup, in which disk acts as a buffer between servers and tape devices. By incorporating disk into the backup process, companies can perform non-disruptive backup and recoveries and produce very fast and efficient disk-to-tape copies (especially useful with today’s LTO4 speeds at 120MB/s).
Stage 3—Using replication adds distance (the remote dimension) to disaster recovery. Replication offers the ability to backup files to remote sites, so companies have a fast, convenient way to protect their data safely offsite.
Stage 4—Adding intelligence enables better data management practices, rather than the strict media management activities used in the past. At multi-terabyte capacities, the only way to deal with backups is doing it smarter. Intelligence incorporated in SRM tools can eliminate zero-value data through techniques such as combining VTL with data deduplication, using CDP, creating disk-based synthetic full backups, or even outsourcing backups. And through WAN optimization, companies can now eliminate the need for a physical IT presence at remote offices, either by tying everyone back to the datacenterthrough the network or by remotely managing off-site secondary data copies.
Traditionally, IT administrators implement backup and recovery by installing a backup agent on each server. But in a virtualized environment, you could have as many as 10, 15, or even 20 VMs on a single physical server. To employ this method in a virtualized environment, you’d need to purchase and install a backup agent on each and every VM. Plus, backup applications usually require significant I/O, CPU, and memory resources. Because virtualized servers run at much higher levels of utilization, they have fewercycles available for backup, which can cause unacceptably long backup windows. With these complications combined, the traditional approach doesn’t really make sense in virtualized environments.
To achieve backup and recovery on virtualized servers, you could save a copy of each VM and store it somewhere on the network where you can recover it if needed. In fact, many companies employ this very method, although it does present its own problems. Specifically, you can’t restore individual files without restoring the entire VM. And usually, you don’t want to recover the entire VM—just individual files that need to be reconstituted after someone inadvertently corrupted them.
Virtualization encapsulates everything in the VM, which is part of its allure. For example, you have the freedom to move copies of your VMs between servers that you have scattered around the network. While this mobility is useful in some circumstances, it also presents consistency complications when it comes to backup and recovery.
Our next blog post will discuss 5 basic approaches for data protection in a virtualized environment, so be sure to come back soon, or subscribe to our RSS feed.
