The Buried Lead in Amazon's Elastic MapReduce Announcement

Amazon just announced that you can now run Hadoop based MapReduce operations on massive data collections stored in S3, computed on their EC2 cloud service:

Amazon Elastic MapReduce

This is almost magically beautiful, the possibilities for small science teams to be able to do super-computer level computation (that would traditionally require big teams and big budgets) with a small team and almost no budget... just, wow.

What I also find interesting is the buried lead in this announcement though. That is that this is the first time that Amazon is directly trying to compete against Google's App Engine. The distinction being that App Engine is "platform as service" while AWS is "infrastructure as service". In human words, Google lets you build applications and let worrying about how the infrastructure will scale up your application to meet huge usage patterns up to Google's engineers.

In contrast, Amazon's AWS lets you build your own scalable infrastructure easily with their services, but you need to develop all of your own load balancing and auto-scaling features (or use someone like Scalr's or Rightscale's wrapper around AWS, if you are willing to pay the considerable price)

With this new service launch you just upload your data to their cloud storage (S3), write your data crunching application to their API, then upload it to their service and hit go. This is exactly the workflow of App Engine.

Glad to see the competition is fierce and fast in cloud services right now, this is a great time to be a developer.

Is Salesforce.com 100x as efficient as Amazon?

The Efficient Cloud: All Of Salesforce Runs On Only 1,000 Servers

The raw numbers in this article are pretty interesting:

Salesforce talked about its own back-end infrastructure and revealed that all of Salesforce.com runs on only about 1,000 servers. And that is mirrored, so it is really only 500. Think about that for a minute. Salesforce has more than 55,000 enterprise customers, 1.5 million individual subscribers, 30 million lines of third-party code, and hundreds of terabytes of data all running on 1,000 machines. Amazon's Web Services, in comparison, runs on about 100,000 machines I am told by someone with knowledge of Amazon's server infrastructure.

But I'm a little confused about how he takes those and jumps to the conclusion he leads the story with, that salesforce is 100x as efficient as AWS:

But still, that is roughly a 100 to 1 efficiency advantage that Salesforce has over Amazon's cloud.

In order to make that calculation, it sounds like he's assuming that Salesforce does 1-to-1 the amount of actual work (processing, transfer, etc) as all of amazon's services do. Thats a pretty remarkable proposition to make, and to hinge your entire article on, without providing any numbers to back it up.