In-Depth

Sybase Revamps Large Data Warehouse Solution

IQ designed to be economical, scalable for terabytes of storage

In the market for a very large data warehouse—something in the multi-terabyte range?

Last week, Sybase announced version 12.5 of its IQ data warehouse platform.

The company has successfully sold its very large data warehouse to a variety of high-profile accounts, including the United States Department of Transportation, the Internal Revenue Service, Nielson Media Research, Samsung and other multinational conglomerates.

According to Deborah Harrington, GM of Sybase IQ operations, IQ’s selling point is that it offers a scalable data warehouse architecture at a reasonable price point. “This is a product that’s been designed specifically for data warehousing. We developed two reference architectures with Sun [Microsystems]. That’s Sun’s term for a solution that’s fully designed, pre-tested, pre-built to achieve goals, one of which was to be packaged for very large data warehousing, the other for economical pricing,” she explains.

Harrington says that Sybase, along with Sun Microsystems, sells pre-built, pre-tested data warehouses based on a framework—which Sun calls an Enterprise Data Reference Architecture—that is designed to support very large data warehouses.

Last year, Sybase and Sun built what they claimed was the world’s largest data warehouse—a 49.8 TB behemoth that consisted of 179 billion rows but used only 22 TB of total storage. As a result of its lower maintenance costs, smaller data center footprint, reduced power consumption, and more efficient resource utilization, Sybase and Sun say that customers who implement their Enterprise Data Reference Architecture can save up to $1 million per terabyte of input data.

In the very large data warehouse space—which currently describes a segment of anywhere from 10 GB on up—that can amount to a sizeable cost savings, Harrington maintains. “The kinds of customers that we’re dealing with are really rapidly approaching the 10 to 15 TB limit,” she comments.

The very large data warehouse space has long been the provenance of Teradata, which sells a range of data warehousing and analytic software solutions that are optimized for NCR hardware.

In this respect, says Linda Jarvis, director of engineering with Sybase, even though IQ 12.5 runs on a variety of different platforms—Sun’s Solaris, HP-UX from Hewlett-Packard Co., AIX from IBM Corp., and Windows NT 4.0 and 2000 from Microsoft Corp.—the Enterprise Data Reference Architecture is optimized for Sun hardware. “We’ve certified and tested and developed these packages, we’ve found that we’ve got an exceptional price performance at the low end and at the high-end,” she asserts. “So in that sense, we are competing with the Teradata approach.”

New Analytic Benchmarks

In tandem with the IQ 12.5 announcement, Sybase and Sun trumpeted new TPC-H benchmark results that they say demonstrate the scalability and economical pricing of the revamped data warehouse for more conventional data warehouse configurations.

In the TPC-H benchmarks, the revamped Sybase IQ 12.5—running on Sun hardware—delivered the lowest price-performance (measured in the TPC-H $/QphH metric) in the benchmark’s 100 GB, 300 GB and 1 TB ranges. The Sybase IQ 12.5-powered data warehouse also required the fewest disks to conduct the TPC-H benchmark. At 1 TB, for example, IQ 12.5 required 54 disks; its competitors required 1,263 and 1,408 disk drives. The Sun-Sybase platform typically turned in longer load times than competing solutions, however.

According to Richard Winter, president of large database consultancy Winter Corp., the price-performance of the Sun-Sybase platform “yields low cost systems that are easier and less expensive to maintain” than those of competitors. “The price-performance showing of the Sun-Sybase IQ platform for processing queries at all tested scales of the TPC-H benchmark reveals the strengths of the unique Sybase IQ architecture for analytic query processing and the efficiency and effectiveness of the Sun servers for hosting this database workload,” Winter concludes.

About the Author

Stephen Swoyer is a Nashville, TN-based freelance journalist who writes about technology.

Must Read Articles