May 08 19

Record breaking warehouses break usefulness rules

Posted by Angus Kidman @ 16:18 1 comments

Guinness World Records has some seriously strange entries, including the world's tallest motorcycle, the biggest weight dangled from a swallowed sword and the heaviest lemon (5.265kg, fact fans). It may not be quite as odd as the oldest female tandem parachute jumpers, but the one for the world's largest data warehouse is still pretty bizarre.

The original reason the Guinness Book (as it's apparently no longer known, books being so last millennium or something) was created to settle arguments in pubs.

Even in such hotbeds of IT activity as Silicon Valley and North Ryde, I'm guessing that there's not too many people actively disputing the size of their data warehouses. But that doesn't mean there's not an entry for it.

Having coughed up an undisclosed sum for official certification, Sybase last week pumped out a press release announcing that it was the official technology behind the world's biggest data warehouse. The specs do sound impressive: a whole petabyte (1000 terabytes), including six trillion rows of transactional data and 185 million other data objects. The system, using Sybase's IQ technology and BMMSoft's ServerSM, runs on a Sun SPARC platform.

Naturally, the announcement also spruiked the green credentials of the platform, such details being de rigueur in any enterprise announcement post-An Inconvenient Truth. Speaking of video, if you really want to waste time in the office, you can watch the certificate being handed over by Guinness World Records people here.

There doesn't seem to have been an actual contender for the title beforehand, though Sybase concedes that Wal-Mart's whopping database, which measured somewhere north of half a petabyte, would be a likely candidate.

To my mind, Wal-Mart still ought to have the title in practice, for a simple reason: it's a real-world data warehouse being used to deal with actual business problems.

The Sybase entry is just a show pony. Sure, it's impressive that it contains all that data and can churn through it so quickly, but it doesn't actually do anything other than perhaps persuade enterprise IT managers to consider Sybase when the next integration task comes up. To increase the real-world factor, maybe Sybase could start renting out access to it.

Planet's biggest SaaS project, anyone?