The Day the Cloud Died: Planning for Cloud Failure


Want the latest storage insights?

Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure

Share it on Twitter  
Share it on Facebook  
Share it on Google+
Share it on Linked in  

"The Day the Cloud Died" (sung to the tune of "American Pie")

A long, long time ago
I can still remember how cloud profit used to make me smile
But I knew if I had my chance
That I could make shareholders dance
And maybe they'd be happy for a while

But second quarter made me shiver
All the forward-looking statements I'd delivered
Bad news on the doorstep
Google price-dropped ten percent

I can't remember if I cried
When analyst expectations lied
But something touched me deep inside
The day the cloud died


So bye-bye, the Cloud had gone dry
Drove my margins to the brink and now my board's asking why
And them good old boys spit out the Kool-Aid and cried
Singin' "This'll be the day that Cloud died
This'll be the day that Cloud died."

(Parody lyrics by Rich Brueckner of insideHPC and Dan Olds of Gabriel Consulting. Here’s a podcast discussing the potential for cloud computing failure.)

As you’ll remember from the lyrics above, “The day the music died” is a line in Don Mclean’s 1972 hit song “American Pie.” But cloud users might want to change the lyrics to “the day the cloud died” if cloud computing providers can’t figure out a way to make money.

Reports that even Amazon Web Services is bleeding cash should be enough to make cloud users worry.  We all know that Amazon is a cash machine, but from the analysis, Steve Brazier of Canalys estimates that “Amazon Web Services lost $2 billion in the last four quarters, and the parent is forecasting losses of between $410m and $810m this quarter.”

So let’s assume that these estimates are true, and let’s also assume that since Google and Microsoft do not break down cloud services that it is also true for them. If a company was making money when everyone else wasn’t, they’d be sure to let us know. This, by the way, is no different than what happened to the storage service provider revolution of the late 1990s, but this time was supposed to be different.

So let’s say that Brazier is correct and the music might die for some cloud companies, as it already has for Nirvanix. What should you be doing as part of your cloud design to minimize your risk, other than the obvious answer, which is to not use clouds?

One other thing to think about before even starting down the path of a plan B is what happens to your data if you cannot get it out of a cloud that has gone belly up? Having a detailed understanding of the fine print in the contract might not just be a good idea – it might save your business. Examples could include whether creditors would hold your data hostage.

Who own the rights to your data? Who has the decryption keys? Just some food for thought as you start to think about what to do and what not to do.

Hedging your cloud computing bets

One obvious answer is to put some requirements into your contract for getting your data out via a network, disk, tape archive or something. This of course doesn’t matter if the company goes out of business or files for protection from creditors, nor does requiring the company to keep networks running for X amount of days so you can get your data out.

Nirvanix users were lucky enough that their networks stayed up long enough to get the data out of Nirvanix, as far as I can determine. But what if you have petabytes of data to move out of your cloud? The cloud vendor might have lots of bandwidth, but do they have enough network and storage bandwidth to drive the networks at full rate and get everyone’s data out before they go belly up?

For example the time to read a 6 TB disk drive is about 9 hours and 40 minutes (6TB / 172 MB/sec read rate). That is a long time. Do you have on your end enough bandwidth and the capacity to handle all of your data that you have uploaded over the years?

So it strikes me as pretty obvious that the only way to move off of one cloud provider is to move to another, as they are likely the only ones with the ecosystem ready to allow you to move your data quickly. I think the key thing is you need a clear understanding of your data, call it a triage of you data.

Submit a Comment


People are discussing this article with 0 comment(s)