LAS VEGAS -- IT industry giant IBM plans to roll out NoSQL technology inside of its flagship DB2 database management system early next year, according to Curt Cotner, the company's vice president and chief technology
The move comes in response to growing customer demand for NoSQL technology, said Cotner, who spoke yesterday during a keynote address at IBM's Information On Demand 2011 conference here.
"All of the DB2 and IBM Informix customers will have access to that and it will be part of your existing stack and you won't have to pay extra for it," Cotner said. "We'll put that into our database products because we think that this is [something] that people want from their application programming experience, and it makes sense to put it natively inside of DB2."
NoSQL databases have become increasingly popular in recent years, thanks largely to the technology's promise to help organizations analyze ever-growing data stores more efficiently than traditional relational database management systems.
The name "NoSQL" originally referred to database management systems that had application programming interfaces (APIs) which allowed users to quickly retrieve information without using the highly popular SQL programming language. Users eventually realized that without support for SQL, it was necessary to write a custom program every time they needed to produce a report.
As a result, most NoSQL technology stacks adopted SQL and changed the meaning of the name from "No SQL" to "Not only SQL," Cotner told the crowded room of conference attendees.
Users should prepare for NoSQL learning curve
IBM's plan to roll out NoSQL technology inside of DB2 made sense to conference attendee Gerard Ruppert, an IT consultant with John Daniel Associates in McKees Rocks, Pa.
"I think ultimately [IBM has] to go there because of the size of the data that's moving around nowadays," Ruppert said. "But it's going to be a learning curve for a lot of the midmarket people because they just don't have that expertise yet."
The appeal of NoSQL lies in its ability to handle large volumes of data faster and more efficiently than traditional relational database management systems, according to Ruppert. He advised that before taking advantage of the new technology, organizations should make sure they have the right skills in-house. Those that don't should consider bringing in some outside expertise before things get messed up, he added.
"In our own practice, we often go in and clean up after other people who don't know what they're doing," he said.
NoSQL database management systems have a reputation for helping organizations analyze so-called big data stores. But "the jury is still out" on whether the technology is right for handling transactional systems, such as those used by banks and other institutions to process things like credit card orders, online purchases and stock trades.
"I think that if you asked our database guys, they would say that they're generally not seeing deployments of technology like that for OLTP [online transaction processing] purposes," said Ted Friedman, a data management analyst with Stamford, Conn.-based IT research firm Gartner Inc. "The vast majority of the usage is going in the analytics direction."
Friedman added that IBM's decision to offer NoSQL capabilities is in line with other industry giants who have made Hadoop, NoSQL and big data announcements of late. For example, Oracle yesterday announced the general availability of its new NoSQL database.
"It's consistent with how we see the relational database model evolving over time." he said. "IBM is doing it and others are as well. You saw Oracle at OpenWorld the other week making announcements around Hadoop and NoSQL capabilities and you see Microsoft doing some other things, so it's a really big deal."
How the NoSQL plan came about
IBM already has some experience in the realm of NoSQL technology, according to Cotner. The company's Rational Jazz collaborative software delivery platform uses a "triplestore," much like those used in NoSQL databases. A triplestore allows users to easily store and quickly retrieve metadata and other information, according to Cotner.
IBM's Rational team ultimately found that triple it didn't have the availability characteristics they needed, such as failover support and the ability to scale out to multiple nodes, Cotner said.
"They discovered that if they were ingesting large amounts of triples in a short period of time, the indexing in the NoSQL store would kick in and lock up the database," he explained. "So, we actually took one of these NoSQL triplestores from the open source [community and] we modified it to sit on top of DB2 so that it can use DB2's indexing, DB2's logging, DB2's solution for high availability [and] and all the things you would expect."
Cotner said the modified NoSQL stack ran four times faster with DB2 than it did in its original open source form and the availability and scalability problems were eliminated.
"You'll see in the not too distant future that the Rational team will start re-hosting all of their NoSQL on top of DB2," he said.