Proponents of grid computing tout the technology as the ultimate way to handle bioinformatics data and applications: Researchers will soon be able to enter a query into their local computer, and the grid machinery will invisibly churn it through the best combination of data resources, applications, and computational processes to bring back the proper answer.
That scenario may be a decade or so off, but in the meantime, early adopters of the technology are plugging into the grid a piece at a time. Last week, Aventis became the latest member of the big pharma club to get in the grid, when Avaki announced that the pharma company had deployed its Avaki Data Grid software across multiple research sites in North America.
Other pharmas, like Novartis and Bristol-Myers Squibb, are deploying desktop grid solutions that harvest cycles from otherwise idle machines; while others, like GlaxoSmithKline, Merck, and AstraZeneca, have opted to start with the data and application aspects of grid computing by collaborating on the UK-based MyGrid project.
For Aventis, the approach is also data-centric, as the Avaki software provides federated data access to widely distributed users. “We make all that data readily available to the authorized users of the grid at Aventis, but they never need to know that they’re specifically accessing some piece of information,” said Tim Yeaton, president of Avaki. “It's all presented to them and to the application as if it were local, as if it were theirs.”
The agreement is significant for Avaki, Yeaton said, because “it’s a signal that we’re seeing really strong commercial adoption across a number of pharmaceutical companies.” Avaki already claims Pfizer as a customer, and Yeaton said the company is also working with several other top-tier pharmaceutical companies.
By focusing on data access, Avaki is addressing a facet of grid computing that may offer the most immediate benefit for discovery informatics. Biologists certainly need computational power, but most bioinformatics applications are not that compute-intensive. What biologists and bioinformaticists certainly need is real-time access to a rapidly multiplying set of data sources. According to Avaki, its software, which does not require an underlying compute grid infrastructure, is able to do this in a simple and seamless manner.
Aventis declined to comment on the deal for this article, but Yeaton said the company was dissatisfied with its existing solution for sharing data across research teams — copying data files. Not only was this time consuming, but it also introduced errors, inconsistencies, and synchronization issues between data sets. “In the drug discovery process, any time you have inconsistency in research data or any issues of synchronization, you run a real risk of invalidating that bit of research and slowing down the drug discovery process,” said Yeaton.
While some of Avaki’s pharmaceutical clients are seeking a method for general sharing of file-based data, Aventis was “looking to share data from multiple sites for running research-oriented jobs on some particular system,” Yeaton said.
Aventis implemented a global integration solution several years ago based on IBM’s DiscoveryLink middleware, which also uses a federation-based approach. An Aventis official confirmed last week via e-mail that DiscoveryLink is the company’s “primary” R&D data integration solution, while “Avaki is more of a niche solution for us in the bioinformatics area.”
While unable to provide specifics on how the Avaki and IBM solutions are being used together at Aventis, Yeaton noted that Avaki works “very closely with IBM’s Life Sciences team,” and described the two products as complementary.
By focusing on data access, Avaki is likely to see more success in the grid computing market than vendors targeting the compute infrastructure side of the equation. A recent study conducted by Platform Computing found that social and organizational issues, such as the reluctance to share computational resources, outweighed technology concerns for 89 percent of respondents when considering the adoption of grid or distributed computing technology. While the local IT guy may be guilty of so-called “server hugging,” it’s unlikely that “database hugging” poses a significant threat to the adoption of data grid technology.