In a recent forum discussion on the relevance of Backup Solutions, one poster asserted that backup software wasn’t that important. Having spent the majority of my career working for a Backup, Recovery, and Archiving (BuRA) Vendor, as well as being a current employee of VCE, I could not possibly disagree more with that idea.
A complete backup, recovery, and archival solution is critical for any business of any size.
To the original poster’s credit, he did clarify that modern disaster recovery and high availability solutions are key to maintaining business uptime. However, I wanted to write this post to re-emphasize the importance of backup.
As virtualization admins, it is easy for us to get caught up in vSphere’s great availability & performance management features such as DRS, SDRS, vMotion, SVMotion, HA, FT, etc, and to forget about traditional backup. There is still room for backup in the next generation data center. However, your data protection methods must vary according to the types of data that you are protecting. i.e. there is no “one size fits all” solution.
Not all data are created equal!
You MUST consider multiple factors including your business-mandated Recovery Time Objectives (RTO aka “how quickly can I get back up and running?”), Recovery Point Objectives (RPO aka “how granular are my recovery points and how far back do they go?”), and government compliance regulations (aka “how long do I have to keep data in case of, umm, ethical misunderstandings?”)
Yes, mission-critical systems like e-mail and database servers need to have the highest levels of availability supported by premium solutions such as Continuous Data Protection (CDP)\HA, snapshotting, etc. However, they do still need to be backed up for long-term retention. Also, in the (unlikely) event of corruption of your CDP Journal data, you will want to have a recent backup handy just in case you need to get up and running quickly.
Your backup vendor’s solution should take advantage of vSphere’s APIs for Data Protection (VADP), including Changed Block Tracking (CBT) to perform a granular restore from a backup of your VMDK files (i.e. the ability to restore an entire VM, or individual files on that VM, from a single backup job). Most importantly, your vendor’s solution should be able to facilitate DR Testing in as painless a manner as is possible.
Software upgrades and other changes in the environment are not an excuse to let your data protection processes lapse. Since these environment changes should be part of formal change control events, updating your data recovery points should also be part of the workflow.
Are there any data protection best practices that you find helpful? If so, post them in the comments section below?