The Hoard Facts

Blog archive

No One can Afford this Kind of Data Protection

There's bad news on the data protection front.

Unless you're the hard-working people at Veeam, and you have a plan.

Which, of course, they do, but they're too modest to release it just yet, lest they spoil the experience of people who want to read their new bad news survey, aka their first annual report on the impact of virtualization on data protection strategies.

This independent survey of 500 IT directors conducted by the survey firm, Vanson Bourne, reveals several depressing facts (and I quote from the press release):

  • Recovery of a backed-up VM takes nearly five hours.
  • 47% of such full server recoveries are performed to recover a single file or application item.
  • Nearly two-thirds (63%) or organizations experience problems every month when attempting to recover a server.
  • Failed recoveries cost the average enterprise more than $400,000 every year.
  • Only 2% of all server and VM backups are tested for recoverability each year.
  • 63% of enterprises experience problems every month when attempting to recover a server.
  • Testing recoverability of a single backup takes IT teams approximately 13 hours.
  • A lack of human resources is the top reason (61%) why IT departments do not test the recoverability of more backups.

Reaction: Who's the guy in your company who has to tell the CEO that you blew through $400,000 last year on failed recoveries?

Imagine the scene: The company's top officers are all sitting around the burnished walnut table in the big room with the view. The CEO asks the CIO if he's been staying on plan for the fiscal year. The CIO attempts to clear his throat, but it's like there's a chicken bone in there. He coughs a couple of times in a vain attempt to expectorate.

Finally, he speaks:

"Uh yeah Roger, we've been right on our numbers with one minor exception."

"What's that, Rich?"

"Well, you remember all that neat data protection stuff you sprung for in this year's budget?

"Yes, as I recall Rich, you said that was going to save us beaucoup bucks."

"Right, Rog, that's what I said--and I meant it, I really did--but, uh, as you know, the IT environment is very complex, and we didn't really see just how complex it would become, and how, uh, well expensive it might get, you know?

"No Rich, I don't know. What are you getting at?"

"Well, Rog, one thing led to another and we ended up putting out four hundred grand on failed recoveries--but I can assure you that me and my guys are going to cap that gusher quicker than you can say memory overcommit."

At this point, all eyes are on Rich, who is perspiring heavily and yanking at his collar in an attempt to open his airway. His bald spot is glaringly reflecting the overhead lights, and his lack of pallor is alarming. He seems to be on the cusp of a severe, debilitating stroke.

You get the idea. Rich--who is currently mopping floors at his local Frosty Freeze--gets the idea. All those numbers in the Veeam report are really bad news.

The question is, what are you going to do about them?

Posted by Bruce Hoard on 09/28/2010 at 11:53 AM


Reader Comments:

Wed, Sep 29, 2010

In our non-production environment, we can restore from tape and get a VM back in a few hours, the same or quicker then getting a phyiscal machine up. In our production environment, we can restore from SAN snapshot and get the same VM back in a few minutes.

Wed, Sep 29, 2010 Paul London

$400k AVERAGE - On failed recoveries -really??? Is this down to recoveries of image level backups exclusively? This looks like FUD to me. It would really help if you could qualify the cost breakdown (and products/technologies used!!!) and how Vanson Bourne arrived at such a seemingly abritrary figure!

Tue, Sep 28, 2010 Pete SF

Have you checked out Pancetera? They have a nice technology that optimizes your virtual environments for cases you have mentioned.

Tue, Sep 28, 2010 Brian J. Bartlett Fresno, CA

This is *not* a new problem. I've been using virtualization for a couple of decades now and x86 virtualization almost as soon as it was available (long-time VMWare beta tester). Prior to working De-dupe technology, I used the stop, copy image, archive (compress it), store model. Tedious, yes, but combined with snapshots it worked very well. It also had the side benefit that if an image were hacked, and I knew the point in time of the hack, I could roll-back and lose only some work, not all of it. Also testing of the archive gave assurance that the image was recoverable, which is more than you can say about many backup implementations that take a very long time to tell you that it failed and fail they did, often. Things are different now. You have the ability to do live backups, de-dupe technology has grown up, and hardware is finally coming of age with the advent of SSD's, so you can cobble together a solution. Personally I'd rather take advantage of the appliances and I'd recommend the NexentaStor OpenSolaris/ZFS solution as one approach that will work quite well. There are also other virtual appliance as well in the storage arena, well worth a look. And if your system is based on Hyper-V, Xen or the like, a converter may work although support would remain an issue. Heck, given a nice data-center,a port may be in the offing. Just my $.02. I see a lot of people groping for solutions that already exist.

Add Your Comment:

Your Name:(optional)
Your Email:(optional)
Your Location:(optional)
Comment:
Please type the letters/numbers you see above