I spent most of Saturday July 19 at “Cloud Computing-the New Face of Computing-Promises and Challenges” which I have already blogged about on last Thursday. There was a large turnout for a Saturday morning in July: Cubberley holds about 400 and it was between half and three-quarters full. There were a number of excellent presentations from practitioners in both industry, commercial research labs, and academia. Many of them are now up on http://www.ewh.ieee.org/r6/scv/computer/
The slides from the keynote by Dr. Hamid Pirahesh, an IBM Fellow in their Almaden Research Center, on the “Impact of Cloud Computing on Emerging Software System Architecture and Solutions” are not up yet, which is a shame because it was an excellent overview of cloud computing. One takeaway: Dr. Pirahesh postulated breakpoints in computing architectures and methodologies at 20, 300, 2,000 and 10,000 processors. Certainly the first, and possibly the second are amenable to multi-core techniques, but the last two are going to rely manycore approaches.
This isn’t Web 3.0, this is something else. Mano Marks, the Google Developer Advocate who talked about the Google App engine made an offhand remark that brought me up short: “ten years ago was the late 90’s.” We’ve moved beyond the web boom (and bust) and fiber build out and are getting a glimpse of a new kind of application that I believe will be as transformational as the initial rollout of the Web was. But it will be used to solve different problems.
Several of the speakers talked about using cloud computing models for processing log files of all sorts, in particular web site clickstream logs and system error logs. The Hadoop project is one example of a ground up re-examination of how to leverage a low cost computing infrastructure composed of fast but unreliable (because they are low cost) processing units. Ashish Thusoo’s presentation on how Facebook is using it to track user activity recorded in web logs is a representative example of this new approach to computing.
My challenge is moving beyond my mental map of existing computing paradigms. I blogged last August about Robert Pirsig’s afterword to the 10th anniversary edition of Zen and the Art of Motorcycle Maintenance where he describes the Ancient Greek perception of time. Time carries you on the back of an oxcart, facing the road you have already travelled:
They saw the future as something that came upon them from behind their backs with the past receding away before their eyes. When you think about it, that’s a more accurate metaphor than our present one. Who really can face the future? All you can do is project from the past, even when the past shows that such projections are often wrong. And who really can forget the past? What else is there to know?
It’s difficult navigating by an outdated mental map of a landscape that is undergoing radical transformation. It’s hard to believe but Salesforce is also a decade old. I was disappointed that Jim Rivera, VP of Product Management at Salesforce.com, misjudged his audience and spent an unfortunate 30 minutes giving a sales pitch that stood in contrast to the rest of the speakers on Saturday. He managed to mention “multi-tenant hosting” so many times that I thought I had taken a wrong turn after the coffee break and ended up at a different event. It’s unfortunate when a company misses an opportunity to engage a technical audience as effectively as the other speakers did.
If you missed the event the presentations are quite detailed and worth a look.