Developers are exerting greater influence on new versions of IBM’s DB2 database, according to one of Big Blue's information management strategy chiefs.
Bernie Spang, director of strategy and marketing for database software and systems, said while IBM has historically consulted DBAs on new features they’d like in IBM’s mighty database, that has changed.
In an interview with El Reg, Spang said IBM is now taking a more balanced approach. “It’s not shifted from one to the other, it’s got to be both,” he told us.
Driving the change is a need to make DB2 more comfortable for developers building web and big data apps that suck on the DB2 data store.
Graph stores find connections between data, so you don’t have to search through piles of relational tables or raw info using Hadoop. Graphs are popular with social networks like Facebook and LinkedIn trying to establish connections between individuals on their sites.
Looking ahead, Spang called key-value pairs a “hot area”. This might be one possibility for future inclusion in DB2.
“It’s going to be driven by the application developers, that’s what’s pushing this, which is different historically – historically we’ve spoken to the DBAs,” Spang said of the recent and future changes.
“I got asked in one interview: ‘Did the DBAs ask for the RDF triple store?’ No, the DBAs didn’t ask for that, it’s for those who want a simpler structure to do things in a different way,” the strategy boss said.
“It was same with XML,” he said. IBM announced XML in DB2 10 years ago with XQuery – the programming language to query XML.
Spang continued: “The DBAs said: ‘I don’t want to put XML in my relational database' and we said: 'Right – we are giving you the same software, it’s just a different structure.' It’s a way for the app developers to find ways that are going to be faster, simpler easier for them.”
Support for XML gave DB2 and its associated tools the ability to combine structured and unstructured data, with the ability to search the data’s metadata.
IBM’s feeling its way on new capabilities like RDF data store in DB2.
Graph data is stored using the Resource Definition Framework (RDF) and queried using the SPARQL query language to look for triple patterns, conjunctions, disjunctions and optional patterns.
“We will bring it [RDF] in, see how it goes, and if this turns out to be a relatively small niche or a passing fad, because the next thing comes along and supersedes it, we will evolve,” Spang said.