Amazon had to prepare for peak traffic, but how could they monetize that excess capacity during non-peak times? With the rise of cloud computing, they found their answer, and Amazon Web Services was born.
Its pretty close to an accidental empire, said Dana Gardner, president and principal analyst of research firm Interarbor Solutions.
The excess capacity point is a common misperception and not at all why we started Amazon Web Services, said Kay Kinton, PR manager, Amazon Web Services.
Developers here at Amazon have spent over a decade building out the infrastructure to run our own Web-scale application, Amazon.com. In that time weve learned a great deal about how to operate very efficiently at very high scale. We started Amazon Web Services because we feel that we can provide tremendous value to developers given the experience, expertise, and assets Amazon has acquired over the past fourteen years.
(I dont mean to put Breen and Gardner on the spot here. Both mentioned up front that this was a word-on-the-street sort of story, and they didnt know the particulars. As far as Im concerned, the main difference between the two versions is intention. Accidental or not, the major plot points remain the same.)
Regardless of the creation story, the fact is that Amazons business strategy since getting into cloud computing is anything but accidental.
In this era of Webification, where everything that can move to the Web does and preferably as a service being a cloud provider means many different things to many different people.
What it means to Amazon is, first and foremost, on-demand capacity. Be it server, storage or database capacity, Amazon will give it to you in a cost-effective and flexible manner.
What Amazon does not do is deliver applications. Amazons cloud offering is perfect for developers, less so for everyone else.
Is that, however, a bad thing?
I dont see their approach as a drawback, Gardner said. Compare what Amazon does to SaaS. SaaS is model of delivering applications. As Amazon sees it, cloud computing is about delivering capacity.
To emphasize his point, Gardner compared Amazon to the early days of Microsoft. It may be difficult to remember, but at first Microsoft delivered an operating system and not applications. Lotus Notes, WordPerfect and others pioneered the application space. Only later, after Windows had a firm hold on the operating system market, did Microsoft start to divide and conquer everything else.
The lesson is that you dont want to compete with your customers, at least not at first, Gardner said. Why make a little bit of money on applications when what the market really wants today is cheap capacity?
While Amazon, along with Google, leads the cloud space, lets be clear here: cloud computing is in its infancy. This fact alone prompts skepticism from many analysts.
Amazon needs to decide whether cloud computing is a hobby or a business, said Nik Simpson, an analyst with the Burton Group.
At the moment their entire infrastructure is hosted, I think, in two data centers, one of which is much larger than the other. If one data center goes down, do they have the capacity to handle everything in the other?
As Simpson sees it, while Amazon has a ton of capacity, they may be missing the boat on some of the things that the enterprise demands, such as high availability and failover. For SMBs and developers testing projects, the low cost makes this a reasonable tradeoff. Yet for enterprise customers curious about the cost savings associated with Elastic Compute Cloud (EC2) and Simple Storage Service (S3), the risk may trump the low cost.
Amazon, predictably, believes it is already enterprise ready. Amazon argues that they offer failover capability, noting (in Amazon-speak) that they currently expose 6 different Availability Zones to customers in two different regions.
Amazon also touts its 99.95% uptime SLA, as well as a number of enterprise customers, such as Eli Lilly, the Washington Post, NASDAQ, eHarmony and Virgin Atlantics new travel website, VTravelled.com.