Bench Press

We’ve mentioned before the power of Cloud Computing as a means of expanding the computational power in the hands of researchers and scientists (by “outsourcing” it to someone with excess computing capacity like Amazon or Google), such as those at CERN studying high-energy particle physics. So, it was very heartwarming to see a Chemical & Engineering News cover story on the use of new cloud computing systems by large biotech/pharmaceutical companies like Genentech, Pfizer, and Eli Lilly.

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."

What was most interesting about the article was the assessment of the cost/benefit analysis that the companies each went through before adopting the technology, something which is important to understand both for researchers/companies interested in adopting cloud computing techniques as well as for technologists/developers striving for broad adoption of their own special technologies. This particular case is especially enlightening as it flips the conventional wisdom – who said that all pharma IT systems and managers are slow to embrace change?

There are 4 key factors that seemed to play a role in “tipping” the large bio/pharma companies: (1) cost, (2) maturity, (3) third parties, and (4) the ability to achieve sizable benefits on a number of scales.

1. Cost: This is usually the “easy” part of new technologies. After all, if a technology is being considered as an alternative to conventional approaches, it usually has major cost or efficiency advantages. In this case, Cloud promoters/IT experts were able to make the case that the cost of buying, setting up, qualifying, and supporting IT infrastructure needed to support all internal demand for computing resources was much greater than the cost of outsourcing some computing needs to Cloud Computing providers. This cost, of course, is measured not only in terms of IT dollars spent,  but also in terms of the time needed to set up, qualify, and develop the necessary IT infrastructure as well as the potential dollars a company might lose by being less nimble.
2. Maturity: One of the major obstacles that new technologies face is that nobody trusts them. After all, why trust a new player with an unproven product for your ultra-important needs (especially when your job is on the line if your needs aren’t fulfilled), when your current setup is working “good enough”? In the case of Cloud Computing, however, the availability of well-known vendors (Microsoft, Google, Amazon, Rackspace, etc) who have had years of experience developing their own systems and product offering has been instrumental in wearing down the traditional resistance that new technologies tend to face. Amazon, especially, has built a well-known cloud platform (their Elastic Compute Cloud more commonly referred to as EC2), with:
   • Easy-to-use web interface
   • Range of offerings that are attractive to many customers (e.g. targeted offerings which distinguish the needs of web application providers versus high-performance scientific computing clusters)
   • Support for standard enterprise software to minimize the amount of re-work that CIOs/IT people will need to do
   • Professional SLA (service-level agreement) which legally obligates Amazon to provide a certain level of service
   • Simple billing system
3. Third party: The existence of third party players on a particular type of technology provide two things to companies interested in adopting a new type of technology. First, it is proof of a concept’s maturity. If a technology really is great, then there will be companies that aim to provide services or add-on products for users of that technology. Second, it simplifies the job of using the technology, as there will be companies who specialize in providing support and useful add-ons. In the case of Cloud Computing, there are a number of providers who specialize in helping companies deploy Cloud computing software or manage their use of computing resources. Companies like Eucalyptus and CycleComputing can provide support (for a fee) for companies interested in setting up their own cloud, and open source projects like Hadoop and research papers on how to use Amazon’s EC2 to do proteomics research not only validate Cloud Computing but provide free advice on how to set up research “clouds”!
4. Benefits achievable on multiple scales: Too often, new technologies promise grand benefits which can only be realized on very specific scales. Small pilots and full-scale transformations are not how companies make decisions and do not reflect the reality of how technology is used. Technology that only delivers significant benefits in small pilots or only work if all old technology is replaced by a new one will be very challenging to adopt. The beauty of the Cloud efforts detailed in the article is that they cover a wide range of scales. Benefits are easily realized regardless of if you outsource your entire IT infrastructure, or if you choose to only use the Cloud some of the time, or selectively avoid using the Cloud when regulatory concerns dictate it.

This is not to say that the adoption of cloud computing in these cases went perfectly smoothly. Amazon does experience service outages and there are still issues to be sorted out regarding SLAs and regulatory concerns, but their adoption by companies in an industry known for conservative IT practices is something that can and should be learned from.

(Image credit)