Chances are you have architected your virtual environment to have no single points of failure. (If you haven’t you may want to brush up your resume sooner or later…) Generally we make sure to protect things at various levels in the hardware stack:
- SAN – 2 or more store controllers + RAID + Hot Spare’s etc
- Servers – N + 1 (or more) for capacity + redundant power + redundant NIC’s
- Network – Pair’s of switches + multi-path layouts to protect against switch or Core failures
As well as the software stack:
- create multiple active directory controllers
- Cluster our SQL and Exchange servers
- Use DFS and other tech’s to make file servers highly available.
So my question is:
I ask this because if you haven’t heard VMware’s key component of most virtual backup solutions (Change Block Tracking) has another critical bug that renders incremental backups useless on the vSphere 6 platforms.
(This isn’t the first bug in CBT either… here is another example from 2015 as well.)
Just to be clear, I am not suggesting that data protection products that leverage this VMware API are bad. Nothing even close actually. What I am saying though is that if you have architected your entire solutions to be redundant… I would HIGHLY recommend that you also make sure you are protecting your data with at least two different methods. This gives you some piece of mind when little bugs like this pop up, or when your VMware snapshots go crazy leaving your VM crippled and your backups a mess because VMware was refusing to do snapshots…
Some of the more mature backup products are already taking this approach, they understand there are limitations with VMware’s snapshot technology. Veeam for example… is doing awesome stuff around integrating with SAN vendors and their hardware based snapshots which lesson the reliance on CBT. But there aren’t too many backup products doing that.
Some other options would be:
- Zerto (Disclaimer: I work for Zerto!)
- San based snapshots
- San based replication
So my final thought is that if you are only doing backups one way…and relying on VMware’s CBT then you need to start talking to your technical consultants to find out what else you can do to protect yourself a second way just in case…