Data center horsepower is a serious issue for Powerset, the natural language search engine that plans to launch its service early next year. Powerset will compete with Google and Microsoft, which both wield industrial strength infrastructure. At many points in its process, PowerSet's computation requirements will exceed those of Google and other keyword-based search indices, according to Powerset chief operating officer Steve Newcomb.
Where's all this computing power coming from? Powerset is pursuing a two-pronged strategy, using its own data centers for some tasks while outsourcing much of the intensive indexing work to Amazon's EC2 utility computing service. Powerset is not revealing much about its in-house infrastructure just yet. But it's releasing one of its resource planning tools, which has helped the company plan its data center strategy. The tool, which uses a Flash interface to an Excel spreadsheet, analyzes the most affordable use of computing power, comparing the cost of leasing, buying and using a virtual appliance on a utility computing service.
"We built a spreadsheet to Flash interface so you can embed the presentation and have it ready for executives," said Newcomb, who provides additional detail on his blog.
