Microsoft researchers and colleagues from Bing have been collaborating with others from industry and academia to examine datacenter hardware alternatives, and their work, a project known as Catapult, was presented in Minneapolis on June 16 during the 41st International Symposium on Computer Architecture (ISCA). Confused yet? Hold on. Their paper, titled A Reconfigurable Fabric for Accelerating Large-Scale Datacenter Services, describes an effort to combine programmable hardware and software that uses field-programmable gate arrays (FPGAs) to deliver performance improvements of as much as 95 percent.
Now I know you’re confused – so what exactly does this mean? What Bing, Microsoft, and a group of others are trying to do is figure out how to use the same amount of servers to do double the work, or use half the amount of servers to do the same amount of work that is being done currently. Finally how do you accomplish all of that without exponentially increasing your overhead costs (as you would with a traditional cloud platform).
In the evaluation deployment outlined within the paper, the reconfigurable fabric—interconnected nodes linked by high-bandwidth connections—was tested on a collection of 1,632 servers to measure its efficacy in accelerating the workload of a production web-search service. The results were impressive: a 95 percent improvement in throughput at a latency comparable to a software-only solution. With an increase in power consumption and total per-server cost increase of less than 30 percent, the net results deliver substantial savings and efficiency.
So while this all might sound completely irrelevant to you – it’s really not. See if Bing has 40x more time to compile a relevant search result for you than they currently do – you can be fairly confident that the success of this technology will bring about some new ranking algorithms to Bing search results.