IBM DB2 made a market for the relational database in 1983, but its future is in stream processing and unstructured data management.
In the 1970s, Don Haderle and a team of researchers at IBM began a project that would change the face of the company's database business forever.
The fruits of their labor would come to be known as DB2, and would hit the market in 1983 to become a force for IBM among relational databases.
"The biggest challenge with relational was to make it cost-performance responsible," said Haderle, who led the DB2 project from 1977 to 1998. "That's where we toiled for the first six, seven years or so."
Prior to relational databases, the network and hierarchical database languages were assemblylike, meaning they were very low-level, Haderle explained. The higher-level language of relational improved productivity, he said.
By the time Haderle retired from IBM in 2005, he had become CTO of IBM's information management segment and had watched the DB2 product he helped to establish grow in features and functions. Some two decades after the product hit the market, the demands of the enterprise are very much the same-even if what is needed to meet those demands is very different.
"Robust [in the '70s] meant it had to perform well, and the system had to stay up ... for a shift to a shift and a half," Haderle said. "Today systems have to be available 24 hours a day, seven days a week. Back then they had to be available 10 hours to 12 hours a day, five to six days a week, and that was a monumental hurdle to get the systems to be that responsive."
Today's enterprises want their databases to drive business, said Anant Jhingran, IBM CTO and vice president for Information Management. Speed is key, and in the future, IBM will focus on stream processing inside and around the database, he said.
"What we find is that our clients are actually looking to add on on-the-fly processing to the database capabilities," Jhingran said. "They want to mix and match the persistence of the static information with online revelation. They want to do the modeling of historic information so then they can know what queries to run on the fly ... and that plays into IBM's strength."