Brian Bloom is a staff writer at ComputerWorld Canada. You can find him on Google+. He covers enterprise hardware and software, information architecture and security topics.
Teradata Corp. says its new database offering lets you keep a close eye on the health of your data warehouse. Business leaders can keep their fingers on their data’s pulse. And the software will take its temperature and put it to bed if needed.
Teradata 14, the company’s hybrid row-column database engine, allows for much faster queries thanks to new columnar compression abilities, says Alan Greenspan, product marketing manager at Teradata.
“You can get much better compression than you can in a traditional row store database orientation. The data that you’re looking at and that you’re compressing is very homogenous because you’re only then working on the data for a particular column. So, it’s all the same: it’s all customer number or it’s all sale price or data.”
He adds while Teradata uses block-level compression, which places heavy demands on the CPU, the engine can get around the problem by working smart, not hard. It separates “hot” data from “cold” data — in other words, variability in how often data is queried.
With this “multi-temperature capability,”says Greenspan, “the database becomes intelligent enough to only apply that to cold data, data that’s used only infrequently.”
Greenspan says this is the inherent advantage of Teradata 14, the ability to turn the heat up or down automatically in a data warehouse. For example, data from a current month that is accessed repeatedly and stored uncompressed would be compressed in following months as it grows colder.
“But then, next year it becomes the reference month — 12 months versus last year,” says Greenspan. “We want to uncompress it, but we want to tell the database [to do it]. Well, in a data warehouse environment, it becomes too much manual management and humans can’t keep up with it.”
Tony Baer, principal analyst at the Ovum industry research firm, says competition among database software vendors to develop better and faster business intelligence tools is starting to resemble “an arms race.”
“What Teradata is doing, it’s not bleeding-edge by any stretch or means,” he says, “but it’s definitely staying consistent with the direction of the industry. And it’s certainly a solid advancement for their install base.”
Teradata [NYSE: TDC] might not be “the only game in town,” he adds. “But they’re probably the most experienced one.”
But Baer says it’s hard to pick winners in the big data analytics game just yet. Teradata does have an edge in terms of its longevity in the industry, but the nature of the business has changed so much that adapting quickly to the future is what really counts.
“If you think about it, Teradata invented the whole category of big data like 30 years ago,” he says. “It’s just that what was considered ‘big data’ 30 years ago might be able to fit on your mobile phone today — well, that might be a bit of an exaggeration — but it would definitely fit on your PC, on your laptop.”
He says that the old either-or paradigm of database design (row- or column-based) is rapidly giving way to hybrid models based on existing databases. By contrast, he says, the idea of building new columnar databases from the ground up and marketing them is a “very purist approach” that probably wouldn’t appeal to many companies with large, established data infrastructure.
“I suppose in a perfect world, architecting is always going to be the most efficient way. On the other hand… if [there’s] anything that’s going to be a legacy system it’s a database. When I say ‘legacy,’ I’m talking about something that’s entrenched, which requires specialized knowledge.”
“Chances are, a Teradata user is not necessarily going to want to dump Teradata to get a columnar database unless they want to change out all their DBAs.”
