Hardly a week goes by now where some big IT company isn't announcing a new cloud computing platform. Jumping into clouds seems metaphorically questionable, but a lot of IT firms see large-scale utility computing as the next big thing in computing, and they don't want to be left out. Most recently hopping onto the bandwagon are Verizon, Computer Sciences Corp., and Sun Microsystems -- its second foray into on-demand computing.

Those three companies add to a growing list of cloud providers, including Google, Microsoft, IBM, HP, AT&T, and dozens of smaller players. But HPC users seem to be gravitating toward the 800-pound gorilla in the room -- Amazon and its Elastic Compute Cloud (EC2) offering. Even though EC2 has only been around for three years, it represents the oldest and most established general-purpose cloud computing platform.

In particular, EC2 looks like it's becoming the platform of choice for biotech companies. Our February report on startup Pathwork Diagnostics is an example of a small company using EC2 to offload a cancer tissue analytics application. They cited Amazon's $0.10/CPU-hour cost as the main attraction for outsourcing some of their work. Larger biotech companies are using EC2 as well. An article last week in Chemical & Engineering News by Rick Mullin described how a handful of big pharmaceutical firms are tapping into clouds. Pfizer, Eli Lilly & Co., Johnson & Johnson and Genentech are all looking to offload some of their bioinformatics work onto the cloud. From Mullin's article:

Although Lilly has a sizable installed base of computers, the company's IT infrastructure is operating at full capacity, says Andrew Kaczorek, senior systems analyst for discovery IT. "Because we have hundreds of different users, what we see is spiky utilization," Kaczorek says. "The result is that for days at a time our clusters are at 100% of capacity. This means there are actually scientists who have work to be done that is literally sitting in a queue." Although exact cost savings are difficult to calculate, they are clearly significant, according to Powers and Kaczorek, as are the time savings.

For example, the company was able to rent CPU cycles on EC2 to run a bioinformatics sequencing code on a 64-node EC2 cluster. For a 20 minute run, the cost to Eli Lilly was $6.40. That's hard to beat when compared to the price of maintaining those additional 64 compute nodes on a permanent basis.

For bioscience businesses, the cloud story is especially compelling. Unlike other traditional HPC users like government labs, financial services firms, and oil & gas companies, life sciences came relatively late to the information technology game, so computing know-how and infrastructure at these companies tend to be spread rather thinly (at least relative to, say, a DOE lab). But today biotech companies are fully immersed in and dependent upon information technology, especially high performance computing. Mullin continues:

[T]he rapid creation of life sciences data keeps pointing to the use of cloud computing, and this is especially true in the area of genomics research. Advances in nanoscale and microfluidic chemistry now allow DNA to be monitored on tiny beads by photographic sensors that, according to Chris Dagdigian, principal consultant for the BioTeam, generate TIFF images in collections of up to 800 gigabytes. "This creates a massive data-capture and handling problem," he says. "We are now in an era where instruments that are showing up in very small wet laboratories are capable of producing a terabyte or more of data in a day."

It's conceivable that the drug companies will bypass the large-scale datacenter build-out that occurred in other HPC verticals and move directly to an on-demand computing model. As such, it may serve as a model for how other HPC users, especially smaller organizations and new users with little high-end computing expertise, can get cloud-enabled.

The early experiences by these drug firms also point to how security concerns are holding back more widespread use of cloud computing. In this case, their main concern is protecting their intellectual property and patents, but almost all HPC users (not to mention just everyday enterprise users) have security issues of one sort or another. It's worth noting here that Verizon's new cloud platform offers added security, primary because their cloud runs over their own private network. But they also offer additional security in the form of identity and access management, host intrusion detection, application vulnerability assessment, network application assessment and professional security services. It's not too hard to imagine that computing in the cloud can be made at least as secure as it is behind a local firewall.

For the HPC crowd, the longer term concern is performance. For a good synopsis of this topic take a look at Douglas Eadline's recent article in Linux Magazine and the EC2 benchmarking paper (PDF) he references. The main argument put forth is that running applications directly on top of purpose-built HPC machines is always going to be more efficient than running applications through a bunch of cloud layers on general-purpose platforms. My take on this is that focusing on performance and computing efficiency ignores the more useful (but more slippery) concept of productivity. I've yet to see a research paper look at HPC in the cloud from this perspective.

There are some early attempts to marry cloud computing services with traditional HPC infrastructure. Darkstrand, Nimbus Services, R Systems, Univa UD, and a handful of other companies are on the leading edge of HPC-as-a-service that bring real supercomputers into the mix. Wolfram Research is also developing its own HPC Cloud Service in partnership with Nimbus and R Systems. Whether HPC will be able to carve out its own niche in cloud computing is an open question, but a deeper discussion of this will have to wait until another day.