AI query optimization in IBM's Db2 shows you can teach a tech dinosaur new tricks

The database Big Blue doesn't like to talk about gets makeover in November

IBM likes to be secretive about Db2, the 41-year-old database still used by some of the world's largest banks and retailers, but occasionally something leaks out.

Db2 was first launched in 1983 and is seen as IBM's breakthrough relational database, a relatively new concept at the time. Big Blue had earlier efforts, as did Oracle and Ingres, but IBM's total dominance of the mainframe market at the time soon ensured Db2 would rule the sector.

Db2 comes in various flavors – mainly, a mainframe release for z/OS, currently on v13, and Linux Unix and Windows (commonly LUW), for which v12 is expected to arrive in November.

At a recent International Db2 Users Group (IDUG) conference in North Carolina, IBM said LUW v12 would replace its older Tivoli System Automation for automatic switching of users, applications, and data from one database system to another in a cluster. In v12, this would be replaced with Pacemaker technology for cluster management to help detect component failures and orchestrate necessary failover procedures to minimize interruptions to applications.

It also adds name space separation for multi-tenancy support to create a "logical separation between one or more database schemas."

In keeping with 2024 trends, IBM also emphasized AI integration in its presentation. For LUW v12, IBM promises an AI-powered query optimizer, which will "allow Db2 to continuously learn from customer's queries and achieve up to three times query performance improvement over prior version," according to a presentation slide. Big Blue also promises Db2 will be "infused with GenAI" in some non-specific way.

After attending the conference, Db2 specialist Craig Mullins, president and principal of Mullins Consulting, said users were looking forward to the features in Db2 12.

"Replacing TSA with Pacemaker is long overdue and most DBAs won't miss TSA," he told The Register.

He said that AI optimization would also show up.

"Optimization has always been a strong point of IBM's and the work they've put into the AI optimizer has the potential to revolutionize query performance. Using AI based on models of each organization's workloads should enable query optimization to be specialized for each user profile. Of course, this will likely [require] a long-term adoption and acceptance. At least initially, the AI optimization capability will be off by default, so organizations will have to specifically request it before embarking on the AI query optimization journey."

Other highlights of the conference included enhanced support for AI via Watsonx, a built-in AI in Db2 for in-SQL AI with SQL Data Insights functions, and the availability of Db2 via AWS RDS, the announcement for which The Register covered last year.

IBM might not include Db2 in the growth products it wants to promote, but it is relied on by customers including JP Morgan Chase, HSBC, Bank of America, and Tesco.

Mullins said the interest from the user community remained strong. "I would also say that the average age of the conference attendees is skewing lower, which is a good thing for a DBMS that is over 35 years old itself." ®

More about


Send us news

Other stories you might like