November 15, 2009

Cloud Computing, The Next Evolution

On November 4-5 2009, I attended a good CSC Leading Edge Forum on Cloud Computing.

The kickoff by W. Brain Arthur was a highlight for me (he is the author of The Nature of Technology). He provided an excellent conceptualization of cloud and it’s place in overall technology advancement and body of world innovation. Essentially, he sees cloud in the 2000’s as the next evolution from the Internet in the 1990s. As such, the cloud is computational power in the “virtual world,” providing a host of benefits including easy access, connectivity, and cost efficiency. He sees the cloud coming out of the initial frenzy and into a industry sort-out that will result in a stable build out.

Another great speaker was David Moschelle from CSC who talked about the myriad benefits of moving to cloud such as scalability, self-service, pay as you go, agility, and ability to assemble and disassemble needed components virtually on the fly. With the cloud, we no longer have to own the computing means of production.

Of course, we also talked about the challenges, particularly security. Another speaker also spoke about the latency issues on the WAN with cloud, which currently limits some usability for transactional processing.

Over the course of the forum numerous examples of success were given including Bechtel achieving a 90% cost advantage by moving storage to the cloud. Others, such as Amazon were able to put up new web sites in 3 weeks versus 4-6 months previously. Also, Educational Testing Service as another example is using cloud bursting, since they tend to run data center at known cyclical peaks.

Others connected cloud with social computing: “the future of business is collaboration, the future of collaboration is in the cloud.”

In terms of the major types of cloud, I thought the relationship between responsibility and control was interesting. For example:

  • Software as a Service -- more “freedom” from responsibility for service, but less freedom to change service (i.e. less control)
  • Platform as a Service – (Hybrid)
  • Infrastructure as a Service – less freedom from responsibility for actual end-user services, more freedom to change service provision (i.e. more control)

In all cases, the examples demonstrated that organizations do not have a lot of leeway with SLAs with cloud providers. It’s pretty much a take it or leave it proposition. With liability to the vendor for an outage being limited to basically the cost of the service, not the cost of lost business or organizational downtime. However, it was noted that these mega-vendors providing cloud services probably have a better availability and security than it’s customers could have on their own. In other words, an outage or security breach will either way cost, and where is there a greater chance of this happening?

Sort of a good summary was this: “Leading companies are moving development/test and disaster recovery to the cloud,” but then this will reverse and companies will actually move their production in the cloud and provide mainly a back up and recovery capability in house. This is similar to how we handle energy now, were we get our electricity from the utilities, but have a back-up generator should the lights go dark.


No comments: