So another service provider crash has resulted in downtime and lost data. Amazon’s cloud crash has brought down any number of websites, and reminded me of the Gmail crash that happened less than two months ago and which I blogged about here.
You can draw many lessons from this and maybe it makes you queasy about outsourcing your data to outside providers. But I’m more interested in drawing a lesson for those users that still run their own IT and control their own data protection. To me, the question to ask yourself is, “how safe am I really?” To put it another way, “How do I know my data is safe unless I know my data is safe?”
Your answer might be, “I run backups every night and most of the time they complete, so I’m pretty sure the data is there.”
That’s cold comfort, because really, what you have is a Schrödinger’s cat problem: you don’t know if the kitty is alive until you actually open the box and look. And you don’t know your data is successfully protected until you use it to restore.
And that’s where the rubber (or the magnetic tape) meets the road. Restore testing is a pain, plain and simple. This is especially true when you are testing big volumes of data or restoring full systems. And that’s why many of us just don’t do it, or we do it maybe once a year (this doesn’t include those among you who are legally obliged to test restores much more frequently and have a string of working weekends to prove it).
A key part of NetApp Syncsort Integrated Backup (NSB) is that we make testing your backups very easy. There are two methods you can use to verify data quickly. The first is the ability to have instant data access, which simply put is mapping to a backup snapshot (all NSB backups are stored as snapshots). The magic of snapshots is that you can just go look at your data. You can map to a drive in about 15 seconds and poke around in the files, or if it’s a database application you can map the data to a test system and see if the DB services start. It’s all non-intrusive, never touches your production servers or data, and is so easy that even the most junior IT admin can be tasked with testing a representative set of applications every day while he’s sipping his morning coffee. You can see it in action here.
The other way is by leveraging virtual machine technology. Any backup snapshot can be started up as a VMware virtual machine, whether the source is virtual or physical. So junior IT guy can start up specific applications each day as VMs and make sure they boot up successfully. If you pre-configure the jobs, all you have to do is click the little green arrow to run the job. This way, junior can start the jobs, go get his coffee, and verify they worked by the time he gets back. You could schedule it too, if one-click seems too strenuous. You can see Instant Virtualization in action here.
And I should mention that if you are using NSB with replication, you can run all the tests off the remote site copies too.
When it comes to data recovery, uncertainty is the last thing you want to feel. Our old friend Murphy is always lurking around the data center, and sure enough he’ll show up at the worst possible time. Better to move to a solution like NSB, where you can easily know what you need to know, and you can give up those long weekends sitting around testing your disaster recovery.
{ 0 comments }
