AOL Taps Vertica Database

Column-store platform edges out IBM Netezza, Oracle and others to provide self-service data sets for advertizing optimization and internal traffic analysis.
"We've got data, and we're not afraid to use it."

This is how AOL describes the mission of its "armies of researchers, engineers and other thick-framed-glasses types" who deliver sought-after online audiences. To let customers and internal business users do more of this work on their own, AOL has added Vertica to its portfolio of data warehousing tools, it was announced on Tuesday.

Vertica was selected to replace MySQL as the database of choice for delivering online data sets and reporting applications. As these data sets have grown, the company was running into performance problems including data sharding and difficulty replicating information across multiple servers, according to Mark Ettrich, senior technical director at AOL.

"Vertica will help us support robust reporting and data consumption whether it's through front-end Web services, APIs or direct SQL queries against Vertica boxes," Ettrich told InformationWeek.

If your last frame of reference on AOL was the 1998 romantic comedy "You've Got Mail" or even the subsequent flame out and divorce from Time Warner in 2009, it might surprise you to learn that AOL's is said to be the number in network in online advertising, reaching 74 out of comScore's top 100 Web sites. AOL has also morphed from a dial-up Internet access provider into an original content publisher, with rising sites including Engadget, Politics Daily, AOL Music and Black Voices.

AOL will deploy Vertica over the coming year to power both external and internal self-service query and reporting applications, Ettrich said. AOL Advertising and customers, for example, want to measure brand awareness and analyze the success of advertising campaigns. Internal customers might want to examine traffic sources and trends, or revenue scenarios.

Vertica's column-store database supports massively parallel processing (MPP) on commodity hardware. The column-store approach -- an architecture shared by competitors Sybase IQ and ParAccel -- maximizes data compression and speeds analytic queries that typically interrogate only selected dimensions across rows of data. MPP further enhances performance by spreading workloads across tens, hundreds or even thousands of nodes.

AOL was already using (and continues to use) Oracle and Netezza for data warehousing, but the company considered products from Aster Data, Greenplum, Oracle, Netezza and Vertica in the MySQL-replacement evaluation.

AOL's requirements included the ability to serve up multiple 10-to-15 terabyte data stores with continuous data loading for near-real-time reporting. The company also tested the ability to support at least 300 concurrent users.

"One of the stand-out features of Vertica that appealed to me was the ability to run on true commodity hardware," Ettrich said. "Some vendors say they run on commodity hardware, but they stipulate particular configurations, interfaces and configurations to get optimum performance."

AOL uses reverse auctions to purchase its hardware, and it specs out several commodity (White) boxes that can be used to run the Vertica database with stand-out performance, Ettrich said.

Asked if AOL intends to consolidate data warehousing platforms, Ettrich demurred, "I don't want to make any statements about that at this point."

Founded in 2005, Vertica was among a handful of upstart analytic database providers that emerged and made inroads against data warehousing incumbents Oracle, IBM, Microsoft and Teradata in recent years. Consolidation in 2010 saw two early pioneers, Netezza and Greenplum, acquired by IBM and EMC, respectively. Aster Data, InfoBright, Kognitio, ParAccel and Vertica are among the independents remaining.

AOL raises Vertica's customer count to 328 firms, according to the company, and it lists Bank of America, BlueCross BlueShield, Comcast, Sunoco, and Verizon among other prominent customers.

Editor's Choice
James M. Connolly, Contributing Editor and Writer
Carrie Pallardy, Contributing Reporter
Roger Burkhardt, Capital Markets Chief Technology Officer, Broadridge Financial Solutions
Shane Snider, Senior Writer, InformationWeek
Sara Peters, Editor-in-Chief, InformationWeek / Network Computing
Brandon Taylor, Digital Editorial Program Manager
Jessica Davis, Senior Editor
John Edwards, Technology Journalist & Author