This two-part video series explores the advantages and pain points of disaster recovery as a service. Part one detailed the positives of cloud-based recovery. Part two provides areas to watch when implementing a cloud recovery service.
Storage analyst George Crump said he is "generally very positive on disaster recovery as a service." But there are elements of a cloud recovery service that you need to keep an eye on.
In his presentation at the Storage Decisions 2016 conference in Chicago, Crump outlined five hidden gotchas of disaster recovery as a service, or DRaaS.
Perhaps the most important issue is identifying how your system will perform after a failure. Users want to be running at full speed, not simply up and running.
"User expectations are higher than ever," Crump said.
In addition, organizations are using an increasing number of applications. Crump noted that he recently spoke to someone at a midsize company who claimed to have 50 mission-critical applications, while the typical amount would be between five and 10.
With the increase in applications, data, expectations and cost, a cloud recovery service may be an organization's best bet. The customer does not have to pay for a full secondary site, and DRaaS is relatively easy to test, which is an important piece of disaster recovery planning that sometimes gets overlooked.
While there are many positives to cloud-based disaster recovery, it's important to pay attention to the technology's challenges. Other possible pain points include:
- networking issues during recovery;
- an unpredictable timeframe for the return to operations;
- overkill for the more likely server or storage disaster; and
- getting data back to your data center.
If you're thinking about implementing a cloud recovery service, watch the video above and read the transcript below for details on the difficulties.
Transcript - Pay attention to cloud recovery service challenges
Editor's note: The following is a transcript of a video clip from George Crump's presentation, "The hidden gotchas of cloud-based disaster recovery and how to fix them," from Storage Decisions 2016 in Chicago. The transcript has been edited for clarity.
There are networking issues that need to be resolved, because their data center is not your data center. Most cloud recovery service providers are actually pretty good about getting you through this. Most of the time, this is resolved.
The return to operations timeframe is incredibly hard to predict. Movement of data is a challenge, and then movement of data for a thousand customers at the exact same time is a challenge. There could be hundreds, if not thousands, of recoveries going on.
It could also be overkill for all the other disasters you're much more likely to face. The chances of you having a site disaster, versus having a storage system failure, versus having an application failure: there's an order of magnitude there. So failing everything to the cloud because an application failed, or your storage system failed, might be overkill.
The big one for us, when we work in these situations, is pinpointing what performance is going to look like. What is performance going to look like in my recovered state? Because people now don't accept the fact that they're just back up and running anymore. They want to be running at full speed.
Then, finally, at some point you will want to go home. Most cloud providers that offer disaster recovery as a service don't want you permanently. If you look at the contracts, generally, there's a point in time, usually about 15 days -- I don't know what's magical about day 16 -- but the cost per compute cycle that you're consuming goes through the roof, which I find hysterical, because these same guys will say, 'Yeah, but you can take as long as you want to recover. Just replicate down.' But a petabyte of data replicating down could take months. They'll say, 'It's OK, you're still running.' No, that would kill me on CPU cycles.
So you've got to get out at some point. This is all really just a quick Band-Aid to get up and running. Many of the cloud recovery service providers have no ability to host you permanently. If you're doing DRaaS into, say, Amazon or Google, potentially you could stay there if you really wanted to. But most people aren't in the cloud for a reason.