I am very intrigued by the following quote from Dave Powers at Eli Lilly, the pharmaceutical giant and makers of Prozac®, Cymbalta®, Cialis® and dozens of other medicines. Their IT challenge was daunting – purchase 64 compute nodes and storage, assemble, and perform their critical testing. The choices were a large capital expense layout with high-cost management or look to a new method to conquer the task. Lilly turned to Cloud Computing and specifically Amazon’s EC2 (Elastic Compute Cloud). Their comment below is the epitome of Cloud Computing. Lilly’s project is fairly typical of geo-science and manufacturing companies that use a sort of ‘modeling’ or simulation approach to discovery, forecasting, and product improvement/ testing.
“Lilly has demonstrated the viability of cloud computing in pharmaceutical R&D, according to Dave Powers, the firm’s associate information consultant for discovery IT. "We were recently able to launch a 64-machine cluster computer working on bioinformatics sequence information, complete the work, and shut it down in 20 minutes," he says, describing a project the firm executed using Amazon’s Elastic Compute Cloud (EC2) service. "It cost $6.40. To do that internally—to go from nothing to getting a 64-machine cluster installed and qualified—is a 12-week process."
(Original article - http://pubs.acs.org/cen/coverstory/87/8721cover.html)
To me, this is the true beauty of Cloud Computing.
5 years ago, I remember a similar discussion about temporary processing projects and implications with an oil and gas executive. He mentioned that the key to their exploration efforts was gathering a bunch of data with ground penetrating machinery. Their machines would “thump” the ground from multiple triangulated locations and gather the reflections from the geologic formations underground. The better the reflections, the more thorough the data but it also meant a few dozen to a few hundred TBs of raw data. The challenge was to process the data as fast as possible such that their team of geologists could find the anomalies and make a decision on the size, location, and viability of extraction of the natural resources they wanted.
The answer 5 years ago for that oil and gas project manager was “renting” a High Performance Computing solution from a notable vendor versus the capital investment to do it themselves. The oil and gas company would have the vendor build the physical 32 or 64 node cluster with storage, deliver the mass of machinery to the oil and gas company site, get the cluster ‘ready’, and then run their data jobs. The company would manipulate the data as required for 2 to 4 weeks depending on how many ‘runs’ they could get processed. In their case, fast was never fast enough. As a “rental”, it was expensive but it was better to “rent” than “own” – the lesser of two evils. After use, the hardware vendor (with their own capital invested) would typically look for another company to coordinate another rental. While some oil and gas companies could justify the investment in large clusters, the processing power and expense was equivalent to chasing something you could never catch. To be more effective over the long term, the oil and gas companies continue to use more sophisticated ground penetrating equipment which gathers more and more data to process. Owning the processing equipment is near insanity from a financial and processing point of view.
Fast-forward 5 years to Eli Lilly. They have discovered Cloud Computing and its benefits. A seamless method to process and extract the information they need. Surely, other industries will gravitate to the Cloud for their modeling and simulation.
That’s the good stuff. What does the Cloud mean to the hardware vendors? If you were a financial analyst tracking a hardware company, what questions do you ask of them?
Q. “Cloud Computing has some great customer benefits but from a hardware point of view, it might mean selling less hardware; assuming you aren’t the cloud vendors hardware of choice, what do you do?”
Possible A. “We are developing newer products that unify compute and storage for more efficient, denser solutions that can serve multiple markets – the end customer that still needs in house control of their transactional business and Cloud vendors that need a massive amount of storage.
While that potential answer is OK, the trend to the Cloud will accelerate. What does it mean to hardware companies that base much of the business on testing, modeling and simulation. A notable vendor once claimed (privately to me) that 50% of their entry level storage products were sold into Enterprise accounts primarily for modeling, simulation, and scaling. That fact seems counter-intuitive to their sales and marketing efforts which geared their high-end products to the Enterprise and their low-end products towards their channel and first-time buyers of that technology.
Assuming customers now have a choice to model their business using a 64 machine cluster for $6.40 in 20 minutes time without the capital expense, the threat is real for hardware manufacturers. The capital investment that larger companies make to model their enterprise will take a chunk out of established revenues of the hardware manufacturers over time by the Cloud. Today that chunk might be small, maybe not. Hardware manufacturers have all faced changes in the market and the good companies know these trends and innovate. The Cloud is the newest challenge and might pose a tougher threat. The Cloud is the first time that end customers have an option of not buying hardware the traditional way and tackling dynamic workloads almost instantly.
The Eli Lilly example solution is a great Cloud example… Imagine an IT meeting with the CIO in a large conference room in Indianapolis, Indiana. The IT department beaming from ear-to-ear with the news that they modeled their new drug in record time with a huge reduction in costs. The CIO impressed with the achievement says, “That is tremendous. What else can the Cloud do for us?”
And thus the Cloud trend accelerates..
