First of all, as I am writing this on this Christmas Eve 2009, and the current year will disappear into history in just a few days, I want to wish each of you the very best, and hope 2010 brings nothing but happiness and success into your lives!
Looking back to years past makes me think about how far we have come in a number of ways, but from a technical perspective, how much of the past is actually circling around into the present. Let me explain by taking you through a short trip back in time.
I began my career in IT back in the 70's after a stint in the US Army. In those days, mainframe was king. Everyone sat around in front of green or amber colored terminals and entered data. (Data entry was actually a great job back then) The move to Mainframe "green screens" was a fairly seamless transition from the typewriters and file cabinets used for data entry and storage back in the 40's, 50's, and 60's.
What I find interesting from a technical perspective as I look back is how IT continually transitions back and forth from centralized model to a distributed model, back to centralized again. When the centralized mainframe model was abandoned for the panacea of distributed computing in the 80's, PC's took hold as the new data entry terminal. Compute power was now closer to the user, and it was a freeing experience not to have to be connected to the network to run an application. As the support costs for this model grew, companies began to realize that perhaps the distributed model was not the panacea it was purported to be, and networked computing came back in vogue (Remember when everyone wanted to become a Microsoft MSCE or Novell Certified Network Engineer?)
Networks continued to expand in the 90's as the Internet took hold, but at the same time, fast Digital Equipment Vax clusters, IBM Mainframe based Linux machines, and enterprise class Sun E10K's with multiple internal domains coupled with centralized Storage Area Network's began to bring everything back into the data center. In the late 90's, companies like Storage Networks Inc and other storage or software service providers (SSP's) came into being, and provided companies with an alternative to buying storage and compute by outsourcing to a third party. So the model was back to distributed. Then the Internet bubble collapsed, and companies brought everything back in house again to the centralized model. (Lack of ubiquitous bandwidth and security were also major concerns back then.)
All this brings us into the present, where cloud computing is the new cool thing, and the distributed computing model is back. There are a couple of major differences between then and now this time, though. This time the distributed model may stick around a while.
The differences are:
- Storage networks are now more secure and distributed with tons of cheap bandwidth.
- Data deduplication has made storage of data less expensive
- Virtualization has commoditized storage and Servers
- Encryption technology has advanced to the point of providing good security
- There are billions of mobile devices which can be used as the new data entry and research terminals.
For the moment, it seems Sun's former CEO Scott McNealy's comment that "it's the network stupid" was right, at least for the time being. We are now nine years past 2001, where my favorite science fiction movie "2001, a Space Odyssey" predicted vastly powerful supercomputers with names like HAL would be controlling our world. (I never did figure out what the whole huge baby thing at the end of the movie was about ...) It's kind of weird that the name HAL is also synonymous with a core part of today's virtualization solutions called the "Hardware Abstraction Layer" (HAL). Coincidence? Something to think about as we celebrate the new year and bring in 2010, which Arthur C. Clark also predicted was the year we make contact.
Merry Christmas, and happy new year!
Christopher Poelker is the author of Storage Area Networks for Dummies, and he is currently the vice president of Enterprise Solutions at FalconStor Software.