Sun, Sybase Save Server Space

A partnership between Sun Microsystems Inc. and Sybase Inc. this week produced their Enterprise Data Reference Architecture, which compresses data to save money (roughly $1 million per terabyte of input data) and create more space in the database. The two companies tested their solution on a 48 TB database warehouse - the largest data warehouse either company has ever created. The Enterprise Data Warehouse Reference Architecture, running on Sun midrange servers with Sun storEdge 9960 systems and Sybase Adaptive Server IQ Multiplex, crammed 48.2 TB of data into 22 TB of total storage. "Most data grows by multiples of five or six using conventional methods, but the data actually shrinks with our solution. Instead of taking up 300 TBs of data we got 28 TB. It saves customers money because it needs fewer servers, less floor space, fewer administrators, and less networking I/O bandwidth. So the overall footprint shrinks considerably to look at the same data. We do all of this without sacrificing the performance," says A.J. Mahajan, senior solutions manager for Sun Microsystems. The space saving solution enables companies to conduct more predictive analysis, instead of being limited to a few months of analysis due to space constraints. The solution also improves query speeds by 10 to 1,000 times. The implementation time has also been crunched. "We can input the data, set up the servers, storage, applications, and the networking in one week. This normally takes at least one month to complete using conventional methods and that does not include the data entry," Mahajan says. "With IQ Multiplex you get an engine that shrinks the data and utilizes the most important features in Sun hardware," says Paul Krneta, Sybase director for the IQ Multiplex and the architect of IQ. "This is one of the biggest announcements for Sybase. We are never in the list of top five big data warehouses. This puts us on the charts. We firmly grabbed the world's largest data warehouse title based on the amount of input data and not storage."