Survey: Biggest Databases Approach 30 Terabytes

Winter's annual survey of the world's biggest databases shows decision support setting the pace as systems double and triple in size. Also a database running on Windows arrives on Winter's top-ten list for the first time.

Led by a surge in the amount of data being analyzed in data warehouses, the worlds largest databases are pushing new heights as they double and triple in size, according to a survey being released on Tuesday.

For the first time in database consulting company Winter Corp.s five surveys of the largest and most heavily-used databases, the largest decision-support database surpassed the biggest transaction-processing database. The flip-flop demonstrates the increasing importance of data analysis for enterprises as they try to better discern trends and patterns, said company President Richard Winter.

"The business drivers are the need to understand customer behavior in a more-detailed way so as to be able to increasingly predict whats going to happen," Winter said. "Doing that well requires increasingly detailed data, and that drives up the size of databases."

For its Top Ten Program, Winter Corp. gathers voluntary submissions from companies worldwide that are running large databases. The program requires that the databases must be in production and contain at least 1 terabyte of data (or 500 megabytes of data if running on Windows). The results, divided into 24 categories, are based on the amount of online data running on the database.

The largest decision-support database in this years survey is from France Telecom and handles 29.2 terabytes of data, triple the size of the top database in that category in Winters last survey in 2001.

Although they were eclipsed by their analytical brethren, the transaction-processing databases were far from laggards. The largest was from the United Kingdoms Land Registry, a government department overseeing land registrations in England and Wales, which reached a size of 18.3 terabytes. That is nearly double the size of the 2001 winner in that category.

Neither France Telecom nor Land Registry could be reached for comment for this story.


For database managers at AT&T Corp., the second-place winner in overall size for decision-support databases at 26.2 terabytes, the more dramatic rise in decision support database size should come as little surprise.

A year ago, AT&T began storing on its Security Call Analysis and Management Platform (SCAMP) database two years of detailed data on calls in its telecommunications network rather, than six months worth of calls. The change occurred because of increasing business requirements to retrieve and analyze more historical information, said Sandy Hall, division manager of real-time customer and service management in AT&T Labs.

"Each record contains 47 fields, so its a massive amount of data and is brought in and kept at the atomic level," she said.

SCAMP runs on AT&Ts Daytona database management system, which it sells commercially. While SCAMP is a decision-support database, it also handles transactions and must be available 24 hours a day, 7 days a week while being able to handle a daily data load of 400 million calls, Hall said.

Next page: Windows databases make the list.