The recent outage at Amazon Web Services coupled with the news that Iron Mountain is exiting some of its storage cloud lines of business has created quite a stir in the storage industry. But many of the conversations in which I have been involved have centered on how some users have been – consciously or unconsciously – applying enterprise expectations to the services that existing cloud storage providers offer. So the questions becomes, “Who is responsible for creating these unrealisticly high expectations – cloud service providers, users or some combination of both?”
When it comes to setting realistic expectations around cloud services, I think the majority of people want to expect the best. They expect the cloud to always work. They expect it to always be “On.” They for the most part do not consider that their data may ever be lost or that one day the service provider will ever say “See ya” or “It’s not our problem.’
But clearly somewhere along the way some users have developed a high opinion of cloud providers. People began to believe that the lights at Amazon and Iron Mountain were always going to be on and that, any time, day or night, they could fire up their applications or access their data at one of these respective cloud providers and access the applications or data they have stored there.
A great example of this can be found on the following thread that was started last Friday on an Amazon forum by a cardiac monitoring company. According to this thread, the lives of hundreds of home-bound cardiac patients that it monitored were at risk because this company was running its application in the Amazon cloud.
Now whether this thread is accurate or not, I don’t know. But it illustrates how apparently at least one company had built its business on the the faulty premise that accessing its servers, data or both in Amazon’s cloud would always be available.
But should this have ever occurred? In other words, why did this company believe for one minute that it could successfully run its business hosting its applications or data with Amazon?
Look at how Amazon Web Services began. Some time ago Amazon looked across its data center and said, “Hmmm, we have a bunch of excess compute and storage capacity. Maybe if we make this excess capacity available over the web at a deep discount price, we can recoup some of our operating costs.”
Nothing wrong with that idea and frankly, I think it was brilliant. But I doubt what Amazon fully expected was how many businesses and people were looking for such a service. Nor do I expect that Amazon expected it to grow so fast and for businesses to start hosting applications like cardiac monitoring in the cloud.
It was just supposed to be cheap, available compute and/or storage capacity accessible over the web. Instead it morphed into something else and while I also suspect Amazon has improved the quality of its data center services, it apparently has not closed the gap.
A similar principle applies to Iron Mountain. Iron Mountain is money generating monster and knows how to turn a profit. Living in Omaha, NE, it was big news over here when Warren Buffett (also of Omaha fame) invested in Iron Mountain because he saw how it operated. It just makes money hand over fist.
So when it got in the cloud services business a few years ago, guess what? It expected to make money. And when it wasn’t making money, or at least as much money as it expected to make, it apparently decided to close the doors to some of its lines of cloud storage business. Steve Duplessie over at ESG made a similar point in a blog entry he wrote back on April 12 where he surmised that Iron Mountain figured out it cannot make any money selling capacity so it got out of the business.
So going back to the question I raised early on in this blog entry about expectations regarding the cloud. Who set these expectations that Amazon and Iron Mountain would always be on and their cloud services would never go away? And maybe more importantly, why did users believe them?
To a certain degree, Amazon and Iron Mountain share some of the blame. As I was writing this blog entry, I was on the Amazon Web Services website and I did not see any big, bold disclaimer saying, “Hey everybody, if you host your applications or data with us, it could all go up in smoke at any moment! So make sure you have a secondary data center to fail everything over and good backups.”
Nor did I see any similar disclaimer on Iron Mountain’s website saying, “Warning! Warning! If Iron Mountain does not make an obscene profit on the data we are storing for you, we are going to shut down some or all of our cloud services without any notice. This will force you to go into crisis mode and scramble to move your data over to a cloud storage provider that is satisfied with lower profits than we are.“
Of course, I would not necessarily expect either of those companies (or any other) to put such disclaimers on the front page (or anywhere) on their website. You might as well close the doors to your business as post language like that.
However because neither they nor any cloud provider post such warnings on their website, users have to be more skeptical than ever. At this early stage of the cloud services game, they need to be willing to do more research than ever to vet out the quality of the provider with which they are storing their applications or data.
Right now it appears some users are consciously or unconsciously granting enterprise level expectations of availability, performance, reliability, support, etc to cloud providers. However as Amazon and Iron Mountain make clear, they have not really proven themselves worthy of being awarded such a distinction for reasons that can now be seen just be looking at the history of these two companies.
So as users go forward with deploying applications and data in the cloud, they should be wary. Cloud providers are not created the same and just because the cloud provider is large or well-known does not mean that it is going to be there when you need them. If anything, these recent episodes suggest exactly the opposite may be true.