Amazon Web Services: A triumph of marketing over reality
Consider this sales pitch: I’m going to offer you cloud services from a big-name provider.
I want to replace your private clouds and perhaps even your datacenters. I’m going to be cost-effective, flexible, and continually expanding my infrastructure to support worldwide cloud integration. And to top it off, the history of my product shows that I’m going to have, on average, at least one major service interruption per quarter that will significantly cripple my services, and therefore your service delivery to employees and customers.
In a piece yesterday on Amazon’s blog post announcement of the advanced features to be included in Amazon Virtual Private Clouds, Larry Dignan commented that, “Amazon Web Services has included virtual private clouds in its EC2 instances in a move that may render the marketing pitches of a lot of hardware companies moot. At the very least, AWS threw a virtual curveball to its physical data center rivals.”
In response, all I can say is that I guarantee you that every hardware vendor in the same space is ready to outline, in detail, every service failure that Amazon has experienced in the last year, and not just the high-profile failures that brought down big name customers like Netflix as recently as 11 weeks ago.
More importantly, those same companies are going to point out that Amazon hasn’t seemed to be learning from the failures, as evidenced by the fact that they continue to happen, and that planning for what currently seem to be inevitable problems means that clients of AWS will need to spend additional monies to cover their own operations when these AWS failures occur.
They can also point at their own business's long history of designing systems which maximize uptime with predictable levels of reliability and effective techniques to address disaster recovery and business continuity.
AWS is clearly a proverbial "game changer" in the delivery of cloud services, and it has definite advantages for certain classes of customers; but until it can show the history of reliability that has traditionally been demanded of high-end datacenters, its success in that market segment remains dependent on strong marketing and a Wizard of Oz-like desire on the part of customers to ignore the man behind the curtain.