Common industry standards - or the lack of them - were the topic of the day when senior representatives from five of the leading database suppliers locked horns in a special debate session at DB Expo in New York this week.
Senior executives from Informix, Sybase, IBM, Computer Associates and Microsoft came together for ?The Great Debate?. Debate chairman Herb Edelstein, president of US analyst group Two Crows Corporation, declared that, assembled on stage, were representatives of all the major database suppliers, a clearly absurd claim given the mysterious absence of market leader Oracle.
Edelstein began by asking which of the panellists were shipping products that fully supported SQL3, the latest version of the SQL database query language. No-one was. Nor could any of them claim to support fully its predecessor, SQL 92. In case any of his audience had failed to get the heavy handed message, Edelstein explained that this highlighted the fact that de jure standards are not necessarily adhered to in practice.
Unfortunately for the ensuing debate the panel was clearly not remotely bothered by this supposed shortcoming. Michael Stonebraker, chief technology officer at Informix, said: "The standards process isn?t working very well. The vendor community is well ahead of the standards community in producing things that are really needed by customers."
"SQL 86 and SQL 89 were market-following standards," he added. "There were different relational products on the market at the time and SQL 86 attempted to resolve their differences." But, he argued, the people involved in defining SQL 92 and SQL3 often have more of a marketing than a technical background, which results in too high level a view of the technology.
Philip Bernstein, repository architect for Microsoft, saw much of the SQL argument as academic anyway. "There are so many issues that people are facing today. Standard SQL is hardly at the top of anyone?s list," he claimed. "People are worried about getting the object relational products they want; they want more flexible data storage; they?re interested in Java and what that has to do to data access. SQL might be on their lists, but I doubt if it gets very much attention."
But if people are interested in, for example, which object-relational strategy to pursue, how confident can they be that they will get some form of standard product? asked Edelstein. After all, the Informix and IBM object-relational models are different, aren?t they?
Not at all, according to Don Haderle, IBM director of strategy and technical planning for database management, there?s a remarkable degree of similarity between them. Maybe so, observed Microsoft?s Bernstein, but two things that are 99 per cent compatible are still incompatible.
Yogesh Gupta, senior vice president of product strategy at Computer Associates, agreed, but saw it as inevitable that there would continue to be different approaches and implementations. What mattered most, he argued, is for users to ask what implementation best enables them to do their jobs and support their business needs.
Having made this valid point, he launched into a shameless plug for pure object oriented databases - such as CA?s Jasmine - before being heckled by his fellow panellists and triggering a new line of debate on the merits of the object-relational model over a pure object model.
Stonebraker, fresh from the launch of Informix? object-relational Universal Server, said he was happy for the market to decide which approach was better. Informix and IBM both had extended object-relational products available. It was up to the customer to try them out and decide which was most appropriate.
Gupta?s response was to dismiss the extended relational paradigm - unfortunately supported by all the other vendors on the panel - as a defensive move. Inevitably there was absolutely no support for such a lone voice in the wilderness and his comments were roundly dismissed.
Fortunately for Gupta, picking on CA is traditionally not nearly as much fun as firing pot shots at Microsoft. It was Informix? Stonebraker who caved in to temptation first with a claim that it "scares the bejesus out of me" that, in his opinion, Microsoft would produce a Java Virtual Machine that was different to Sun?s Java Virtual Machine.
"Microsoft is the only vendor on this stage that is so arrogant that it thinks that it can have its own standards and historically has done so," he said, obviously not noticing the IBM representative to his left. "The thing that scares me is that Microsoft will decide not to commit to SQL3, but will go with Microsoft SQL or Microsoft Java or Microsoft Whatever."
Bernstein was having no truck of that kind of comment. Microsoft had been active over the past six months or so in putting its technology forward to the standards process - which might come as something of a surrprise to the likes of the Object Management Group - but reserved the right to implement standards like SQL in a way that was best suited to its customers, he insisted.
Actually, commented Sybase?s Epstein, sounding rather surprised to find himself coming to the defence of Microsoft, the Java licensing terms are so precise and tight that there is little reason to be worried about what individual licensees are going to do with it.
The panel?s overall conclusion was hardly surprising in a forum that had no end user participation: there are problems with the de jure standards process, but we suppliers are making a pretty good stab at shaping our own de facto standards according to market needs.
"There isn?t going to be an esperanto," concluded Stonebraker. "We?re agreed that SQL3 is the direction in which we?re heading. We?re trying our best. I?m much more worried about the Internet space where we?re all active, but the standards process isn?t even off the mark."
Ceres, located in the asteroid belt, has a carbonaceous-rich upper crust, SwRI study claims
The spacecraft found traces of hydrogen and oxygen molecules, known as hydroxyls, embedded in the rocky surface of the asteroid
The skeleton was unearthed more than 20 years ago in South Africa
Moon's dark side is mountainous, rugged and never visible from the Earth