The pressure on IT departments to be able to store, search and secure ever-larger databases will only increase in the coming year, and so far there are no cheap or easy ways to cope.
"Data volumes are continuing to grow," said Noel Yuhanna, an analyst with Forrester Research Inc. in Cambridge, Mass. "Compliance issues require more data to be kept longer and the usage of unstructured data is growing and making its way into the database. Audio, video, images, faxes are all being added to the database."
To meet the requirements of the Sarbanes-Oxley Act (SOX) and other similar regulations, enterprises end up replicating major parts of their databases. Under the act, employees are allowed to see only the information necessary for their jobs. That often means giving a department its own copy of the database, tailored to include only the data relevant to its function.
"It used to be that you could put everything into one giant database," said Chuck Ballinger, information analyst for Spokane, Wash.-based energy company Avista Corp. "But now we have to divide it and audit who has access to what. We tend to have to replicate it to maintain the separation."
Another data-expanding aspect of SOX requires enterprises to keep far more types of information, including e-mail, electronic transactions with customers and audit logs of who has had access to various parts of the database. And most of those records must be kept for at least seven years.
"We're having to archive a million e-mail messages a month because we're in the energy industry," Ballinger said. "In this environment, there's the suspicion that we could all be Enrons, and regulators may want to search every e-mail. Fortunately, we have all that information archived."
Keeping – and tracking – all the additional data is requiring major outlays. "We bought the auditing software," said Delton Blackwell, design process engineer for InPhonic Inc., a Washington, D.C.-based wireless services company, "and that's going to make us purchase new hardware."
For Ballinger, getting into compliance with SOX "ate the IT budget and all the resources" and will continue to get budget and personnel until implementation is complete, sometime in first quarter 2006. "A lot of projects were put on hold so we could get in compliance," he said.
But even without SOX, databases have been growing. Avista keeps 30 to 35 terabytes of data in its Spokane server room, plus another 3 to 6 terabytes of customer data on a mainframe in Sacramento, Calif.
Avista's database is so vast, according to Ballinger, that it can no longer safely be backed up to tape overnight. That's prompted a switch to removable-drive disk. "The entire system was too slow," he said. "It was taking six to eight hours to do the backup, and we couldn't shut the database down long enough. Our window for error is only a half-hour. If anything happens, we lose the whole day's archive."
If nothing else, database proliferation may hasten the day when tape is obsolete. "When we were working with gigabytes, tape was fine," Ballinger added, "but now that we're up in terabytes, physically, you can't run the mechanism fast enough."