There is a small set of key IT concepts and they do recycle over time
Jim is currently in San Jose and was struck by the fact that he can walk into almost any new startup and run into several people that he knew years ago when one or both of them worked for other companies. One could come away from this type of experience and conclude that there really are only a small number of people who work in IT and they just recycle over time. That conclusion would be a bit fanciful. However, as we will discuss in the next two newsletters, it is not fanciful to deduce that there is a small set of key IT concepts and they do recycle over time.
Jim is currently in San Jose and was struck by the fact that he can walk into almost any new start-up and run into several people that he knew years ago when one or both of them worked for other companies. One could come away from this type of experience and conclude that there really are only a small number of people who work in IT and they just recycle over time. That conclusion would be a bit fanciful. However, as we will discuss in the next two newsletters, it is not fanciful to deduce that there is a small set of key IT concepts and they do recycle over time.
One such concept is centralization vs. decentralization. As an industry we go through periods in which we have a centralized model of IT and later we go through a period in which we have a decentralized model of IT. For example, the IBM mainframe heyday of the 70s and 80s marked an era of centralized computing. Then, driven first by the deployment of mini computers into branch office and later by the deployment of PCs and servers, we entered an era of decentralized computing. The last few years have seen a broad based movement to consolidate at least some servers out of branch offices and place them into centralized data centers. This consolidation combined with the deployment of blade servers has driven a number of folks to talk about “the return of the mainframe”.
Another way to look the dynamic tension between centralization and decentralization is relative to Internet access. Today most IT organizations have a centralized approach to Internet access. As part of that approach, Internet traffic to and from a branch office transits the company’s WAN between the branch office and a centralized point such as a data center. The movement to use more public cloud computing services such as Software-as-a-Service (SaaS) and Infrastructure-as-a-Service (IaaS) will drive more Internet traffic and this will cause some IT organizations to consider deploying a more decentralized approach to Internet access.
We come across a lot of people who tend to dismiss cloud computing because they think that cloud computing is a bunch of rehashed ideas. We certainly agree that many of the cloud computing concepts have been around the block at least once. However, most IT concepts recycle over time and in the case of cloud computing, there are also a few new concepts in the mix.
As noted, most of the concepts that underlie cloud computing are not new. Buying computing cycles from a third party (currently referred to as Infrastructure as a Service [IaaS]) was first done in the 1960s and was called time-sharing. While there is not a difference between time-sharing and IaaS from a conceptual perspective, there is a critical difference from a cost perspective. It was common in the 70s for it to cost between $300 and $400 an hour to rent an hour of time on an IBM mainframe. Today, you can go to a vendor such as Rackspace and rent a virtual machine for an hour and the cost would be 10 cents.
We will continue the discussion of both key IT concepts and cloud computing in the next WAN newsletter. In the mean time, the following URLs point to a couple of exceptionally lucid and insightful discussions of cloud computing and the associated risks.
