Sun Microsystems has teamed up with Greenplum, an open source database supplier, to produce a data warehouse appliance with up to 100 terabytes of storage.
Greenplum is the supplier of Bizgres MPP, a massively parallel implementation of the PostgreSQL open source database system. Frontier Airlines, the second largest carrier at Denver International Airport, uses Bizgres for competitive flight pricing. Bizgres, like PostgreSQL, is an open source code project. Greenplum sponsors the Bizgres project and sells technical support and services on top of it.
Sun officials are letting customers know that if they combine Bizgres with Sun's recent "Thumper" server, the result is an instant data warehouse. Thumper is a tight combination of server hardware and up to 24 terabytes of disk space assembled into a rack mount unit that takes four rack slots—in other words, is about 7 inches high. Thumper is also known as the Sun Fire X4500. The design was produced by Sun co-founder Andy Bechtolsheim, who returned to the company in 2004 to help Sun out of its product doldrums.
Sun in effect has found a vertical application for its Thumper server in the data warehousing appliance.
The appliance can be stacked with 10, 40, or 100 terabytes of disk. At the high end, the price is $15,000 per usable terabyte; at the low, 10-terabyte level, it is priced at $25,000 a terabyte. The high-end system has a higher total price tag, but the price per terabyte has shrunk compared to the low-end system.
The Sun appliance works along similar parallel principles to Wal-Mart's central data warehouse based on a massively parallel system from Teradata, a unit of NCR Corp. The Wal-Mart system allows any Wal-Mart store manager to query a central data warehouse and analyze the day's sales for his store.
Parallel processing allows large databases to be subdivided into sections, with a processor executing a query against part of the database in coordination with other processors. The results are combined to produce an answer to the query that's been quickly executed against the whole database.
The Sun Fire X4500 is running on two dual-core AMD Opteron processors. "The combination drives toward using commodity technology to produce a large, general purpose computing system," said Luke Lonergan, co-founder and CTO of Greenplum in San Mateo, Calif.
The data warehouse appliance is not intended as a transaction processing system for frequently changing real-time data. Rather, it is aimed at retrieval and analysis of large amounts of historical or other static information.
The appliance also uses Sun's Solaris 10 and the 128-bit ZFS file system to address the large amounts of storage available.
The system allows the 30-employee Greenplum, a startup, to walk into enterprise accounts and start talking data warehousing, Lonergan said.