Thursday, October 22, 2009

EC2 Value Shines at Amazon

On the heels of Randy Bias' excellent analysis of the market adoption of EC2 (well 3 wks later but I only read it this week), I thought I would publish the findings of the survey that we conducted on AWS value. While we do not have a huge sample size for the survey (24 responses), I do believe the answers provide some insight into the terrific uptake that Randy describes.

The large majority of respondents (92%) identified themselves as providers of some type of technology application as opposed to enterprise users. I think this mostly reflects these folks friendliness to survey requests versus enterprise users – not a lack of enterprise consumption of AWS. Those in the market with services are more likely to answer surveys due to their empathy for the pursuit of market information. Enterprise users typically have little empathy for the pursuit of information that makes them easier targets for marketers such as myself.

Almost all identified themselves as senior management in their organizations (61%), with 9% claiming middle management and the remaining 30% “breaking their back for the man.” Interestingly, the distribution of the AWS experience curve was not as skewed to the near term as I would have expected. Typically, for a hot/new service, you would expect the majority of the users to be early in their experiences. I consider early to be 6 months to a year. For our respondents, 50% had been using the service for more than a year, with the remaining 50% split 10/20/20 at the 3 month, 6 month, and 12 month experience intervals. I would have anticipated a curve more skewed to 3 to 6 months.

The most popular service of the five we surveyed is S3 (92%), with EC2 (88%) just behind. EBS trailed at 58%, with SimpleDB and SQS bringing up the rear with 17% and 21% respectively of respondents indicated they use these services. Since every EC2 user must use S3, I find the popularity of EC2 to be the most interesting, but not surprising, finding in the survey. It supports Randy's analysis, and it reflects the market generally. The amount of compute cycles sold in the form of hardware on a dollar value basis far outstrips the amount of storage sold on a dollar basis. Also, while there are several storage offerings in the market relative to S3 functionality, few providers have cracked the code on providing compute cycles via a web API with hourly granularity and massive scale in the manner of EC2. EC2 is delivering the big value to Amazon within the AWS portfolio.

To summarize the rest of the responses, scalability was the most important competitive feature followed closely by low cost and pay-as-you-go pricing. Content delivery applications were the most popular workload, with no clear-cut number 2 coming close. Users are spending between $100 and $1000 per month and almost all (67%) plan to add more workloads in the future. Many would like to see an API-compatible AWS capability running on other networks ranging from “my laptop” to their private network to service providers that might be competitive with Amazon. Check out the details for yourself.

My bottom line on this and other indicators in the market is that Amazon's approach to IaaS is effectively the modern datacenter architecture. The market growth of cloud services for compute, storage, messaging, database, etc. will largely reflect the current market for those capabilities as represented by respective sales of existing hardware and licensed solutions. But the availability of these lower friction “virtual” versions of hardware and licensed software will dramatically increase the total market for technology by eliminating the hidden, but real, costs of procurement inefficiencies. When services similar to EC2 run on every corporate and service provider network, we will have more computing and more value from computing. And the world will be a better place.