IBM on Friday will detail a storage architecture dubbed the General Parallel File System-Shared Nothing Cluster that's designed to improve analytics and keep crunching numbers even through failures and malfunctions.
Unveiled at the Supercomputing 2010 conference in New Orleans, the General Parallel File System-Shared Nothing Cluster, or GPFS-SNC, is able to convert terabytes of data into something that's actionable twice as fast as before. The architecture won the storage challenge at the conference.
The gist of the architecture goes like this:
- Each node is independent and self-sufficient;
- Tasks are divvyed up between computers;
- The distributed approach eliminates bandwidth bottlenecks in storage area networks;
- The system can provide data access through outages.
The architecture is a fairly simple set-up that could have implications for enterprise customers.