Serious scaling for open source databases

One of the biggest problems for those who want open source databasing is scaling.

And now it may be solved.

Once you start storing terabytes, and want to do deep business intelligence analysis on that data, a simple mySQL installation on a Dell box is not going to cut it. Proprietary companies have long taken advantage of this, warning firms which hoped to grow fast that they would hit a wall without the right solution.

Greenplum says it has the answer. It's a proprietary layer on an open source PostgreSQL database, enabling true enterprise-class database support with an open source base.

Greenplum president Scott Yara is not blowing smoke here. He has a reference client for this solution, Frontier Airlines. He estimates Frontier is gaining $4-8 million in extra revenue per quarter from this.

"They take data from their core registration system and analyze it to optimize routes and fares." But they couldn't upgrade to a Teradata data warehouse, as competitors had, once they got to 800 million records. So CIO Bob RathRapp looked to Greenplum's "share nothing" system.

Greenplum partnered with a business analytics consultant, DaxbyDaxpy, to come up with something that could run on a collection of low-cost servers. It wasn't cheap, but the underlying technology was still PostgreSQL.

All this works because Greenplum works directly with the Bizgres community, which applies PostgreSQL to BI problems. "We go through great effort to make sure we in the commercial version have full compatibility with PostgreSQL. So when they get people trained they are using Postgre. When people buy into us they buy into Postgre."

So, is the open source database problem solved? Yara thinks so. "We're on a real tipping point," he says.

Maybe. Just don't expect to see Michael Hayden wearing a Tux t-shirt any time soon.