Life Sciences, Analytics and Smarter Infrastructure
Years ago we were very involved with cutting edge life sciences firms doing genomics and proteomics research - cutting edge stuff that was going to help pharmaceutical firms and healthcare in general. Craig Venter and Celera had mapped the human genome and the race was on... (among other things, search for smarter infrastructure that improved results a leg of that race).
We were building out high performance compute (HPC) clusters that dramatically improved searches to identify proteins and their post-translational modifications in samples that had been run through mass spectrometry instruments that were increasing in sensitivity and scan speeds.
In short, they were cranking out a ton of data that needed to be analyzed - back then, the typical analysis could run several days. Using HPC we were able to do the same searches in hours but this required lashing together hardware, storage and networking and distributing the workloads with a high degree of parallelism. I found some links to spec sheets that are the 'productized' versions of what we built with IBM and Thermo Electron: IBM / Thermo BC JS Cluster and IBM / Thermo x1350 Cluster.
One of the things I remember distinctly is that working with the primary investigators (PI) was so different from information technology professionals - the PI wanted to make break-through discoveries driven largely by query and analysis and the IT folks were too busy paying attention to 'traditional' application stuff.
The corollary with the current situation is that smart people driving innovation in organizations today, and the smart IT people are leading different lives. Information technology is still considered a back office cost center, with complexity driving more cost year over year and being forced to do more with less - a less than virtuous cycle particularly if you are of the mind that information technology is at an inflection point where it can become an agent for business transformation and competitive differentiation driven by analytics and optimization. The inflection point I'm referring to is a generally smarter infrastructure - much of it coming from the developments in the worlds of Cloud, DevOps and Analytics.
The good news is that a number of solutions are coming to market that are workload optimized, highly virtualized and componentized. The next evolution is making these building blocks more intelligent and dynamic so that, as workload characteristics change, the eco-system of components adapts within the constraints of policies. Oracle Exa**** and VCE Vblock have made strides in pulling the pieces together. IBM will be announcing their entry into this space and what I've seen so far makes me think it may very well deliver on a progression of this theme:
an integrated system that is modular yet tightly integrated with components that are tuned with and for software
deliver a simplified IT experience - from initial setup and deployment through day to day operations and updates
It's evident that IBM has invested billions of dollars in research to create smarter computers, computers that have become cognitive, more intelligent, more insight driven. IBM Watson provides a glimpse into an entirely new realm of information systems in the future: the learning systems.
On April 11, IBM will unveil a new family of “expert integrated systems” at a special launch event. We'll be keeping a close eye on this one and will report back as soon as we have real world experience... Here is an IBM executive perspective (video):
I would like to acknowledge my friend and colleague Austin Walsh for his leadership and of the SEQUEST HPC inititiave along with the significant contributions of Anatol Blass, Jim Shofstahl and Amy Zumwalt.