Spouting Shite

Microsoft miss the ship(ping container)

Google is the undoubted leader in the scalability race. They run one of the biggest networks of servers in the world, perhaps more than the US government host. They have pushed the edges of traditional technology and invented many, many new technologies, some of which they have released and some of them they have kept secret. Regardless, they are the true scalability kings with a network of servers to rival anybody in the commercial market.

Amazon isn’t too shabby either. They are the market leaders (in terms of $ income) for cloud based services such as Compute (EC2) and file storage (S3). They host a network in several countries (most recently a new addition in Ireland). While Amazon is quite secretive of their network infrastructure you can assume that it is rather large and involves a significant investment to rival the GDP of a small country.

Microsoft is rather different. In pure commercial terms and perhaps technology, they are quite a bit behind the ball. They have no commercially released cloud computing services and no answer to the massive infrastructure of Google or Amazon. It seems that while Google and Amazon were pushing the bounds of the market that Microsoft were involved in some interesting research in both OS (Azure) and DC (containerised DC’s).

It’s generally accepted that Azure is a welcome addition to the market. This OS will bring many more developers into the market, especially those who were perhaps isolated with the lack of a Microsoft OS in the cloud computing market. It will allow easy integration into the technology that many companies already have and will perhaps be an easy way to provide additional premium services to many Microsoft clients. But, without bashing Microsoft too much, it is unlikely that this OS will be ‘market ready’ for quite a while, perhaps a few years. In the meantime it is unlikely that Microsoft will launch anything more than perhaps Hosted Exchange or Hosted SQL on the market. This will be devastating as Google and Amazon will use this time to their advantage…making it very hard for Microsoft to catch up.

The one place where Microsoft is building huge value is in DC design. They have been running an internal research project for the past few years on extremely modular and scalable containerised DC’s that can be deployed in a few months rather than years. This research is being spear headed by a team under James Hamilton. Microsoft partnered with DELL on the designs, which is odd as SUN would have seemed the perfect partner after all the work that they did on Project Caroline.

While the idea of a containerised DC is not new, it is generally accepted that Microsoft (under James) has brought the technology from a simple idea to a near close commercial reality. You can find more information about the teams work here:

·         A video describing the Microsoft containerised DC concept

The idea of hosting your infrastructure in a 40 foot shipping container would be foreign to most IT directors and would get some fired if they presented such a plan to their Board. A shipping container is not as secure, reliable or clean as a traditional DC. At the same time, there are some significant advantages if you are prepared to wear a coat when re-booting your servers. Firstly, the speed of deployment is measured in months and not years, the cost is in $millions and not tens of $millions and the running costs are much lower.

Things were looking up for  Microsoft…until James left his research post and went to work for Amazon. Let’s not assume that the whole project was on James’s head, and that he would be silly enough to take any proprietary intellectual property with him. Rather, the project will continue in his absence….but ALL of the lessons will walk out the door to the market competitor, significantly lowering their costs for building a similar infrastructure. Ouch…that was a kick in the balls.

Only time will tell if anybody will actually deploy a containerised DC. There are still some rather large technical challenges, but they are all worth overcoming in a market when infrastructure is becoming a commodity.