From GigaOm by Patrick Baillie: 'Last year’s fairly significant — AWS outage highlighted the challenges that delivering consistent data center uptime presents. The ongoing challenge of keeping a data center online is a highly complex and often underestimated task, but one that provides the bedrock of any public cloud availability. If the data center fails, the cloud will be offline, and a cloud is only as good as the data center in which it resides.' Continued.
Trish's Comments: I found this post quite confusing. I understand that it is written by the CEO of a successful IaaS company and he is using this opportunity to explain his business model and highlight its benefits. The bottom line is: If you are an IaaS company, you are mainly a software company, you know about networking and software development and management . Don't divide your resources between your core competency and running a data center, which requires a totally different set of skills and capabilities. That is precisely the model that CloudSigma, Baillie's company, has followed and, at first glance, I found his position reasonable and, probably, correct.
Then I started to follow some of the links in the article which help educate us about how Google, Facebook and Rackspace are companies that have followed the opposite model and have been wildly successful. And not only are these companies running cloud services and maintaining their own data centers, they are going one step beyond and getting down and dirty with the design and definition of the servers themselves.
1) In the case of Facebook and Rackspace via The Open Compute Initiative. This initiative was launched by Facebook and has been backed by Rackspace, and other technology giants, since it launched on April of last year. It introduces open source servers that are expected to be 38% more power efficient and cost 24% less to make than current designs. How is that for having knowledge and resources spanning hardware and software at multiple levels?
2) The same can be said about Google. Although it was not of it initial members, it has now been involved for a few months with Facebook's initiative. In addition, it has historically been one of the most innovative companies when it comes to creating its own efficient designs for its data centers and it could be considered as a company that has paved the way for The Open Compute Initiative to launch.
Granted, size matters and all three companies are forces to be reckoned with but it is still interesting to see software companies successfully leading efforts on the improvement of server design.
So, why did I say that I found this post confusing? Every link in his article provides examples against the point that he is trying to make: data center management and cloud services should be ran by two separate companies. I think it is fantastic that he is open and transparent enough as to refer his readers to interesting information even if not in-line with his opinion. At the same time, wasn't this information enough to make him consider that both options are good and that it is the company's specific situation - size, location, target markets - that determines which option might prove most beneficial?