AWS Cloud Pact Shows How Far Teradata Has Come
Next year, Teradata will offer its eponymous database as a service on the Amazon Web Services cloud. It will be the first foray onto a public cloud for Teradata, which has come a long way from its proprietary hardware roots.
It wasn’t that long ago that all Teradata (NYSE: TDC) deployments required an investment in proprietary hardware by the customer. The company sold its flagship Teradata Database—desired by large companies for its fast analytic processing on large datasets—as a pre-configured appliance. At first, those appliances were chockfull of proprietary components, including storage and networking cards.
“When we started 30 years ago, the entire hardware platform was proprietary,” says Teradata Vice President of Product and Services Marketing Chris Twogood. “We had what we called AMPS [Access Module Processors] and they were physical cards that fit inside the physical hardware. We had a BYNET, a physical hardware component that was the interconnect, and it was proprietary.”
Like many other systems vendors, the analytics giant went the proprietary route because it was the only way to deliver the performance that its Fortune 100 clients demanded. But as Moore’s Law progressed and industry-standard technology improved, the performance advantage of proprietary hardware shrank to the point where it didn’t make any sense to go that route anymore.
“Over the years, we’ve taken what was proprietary hardware and converted that value-add into software,” Twogood tells Datanami. “As an example today, AMPS today are all virtualized in software. The last proprietary hardware component that came out of our system just came out a couple of years ago, was the BYNET. The BYNET hardware was converted into software, and we run it on either Ethernet or InfiniBand.”
While Teradata doesn’t yet sell the Teradata Database as a shrink-wrapped software package that customers can run on their own hardware, it has yielded to the call of standards-based technology in the development of its own appliances. Today’s Teradata appliances feature standard Mellanox (NASDAQ: MLNX) InfiniBand interconnects, hard disks from Dot Hill Systems (NASDAQ: HILL) or NetApp (NASDAQ: NTAP), Intel (NASDAQ: INTC) chipsets, and X64 server nodes from Dell. “We configure it, stack it, rack it, install the software, and it’s ready to run,” Twogood says.
Today Teradata announced the next step in that journey to standardization: running on Amazon (NASDAQ: AMZM) Web Services. “The market continues to want appliances, but they also want the flexibility of the pay-as-you-go model, to be able to do hourly payments that a cloud model affords,” Twogood says. “By taking a Teradata database to AWS, we offer that kind of flexibility and choice to our customers.”
The AWS option will be particularly well received by customers who want to analyze data that already exists in AWS, Twogood says. “People want to do analytics where the data sits,” he says. “We often talk about this as data gravity. If your data is sitiing in the cloud, do your analytics in the cloud. If your data is sitting on premise, then do your analytics on premise. A lot of our customers will end up doing a hybrid.”
Teradata has offered its analytic database on its own private cloud since late 2013, but this will be the first time Teradata has embraced a public cloud. Many details–such as pricing, which is a big one for Teradata customers—have yet to be worked out, but the companies have about three months before the Teradata cloud service goes live on AWS to figure that out.
Appliances are still critical to Teradata’s strategy. “Our appliances will deliver the utmost in performance and service level agreements–more so than the cloud will,” Twogood says. “But sometimes people don’t need that when deploying analytics. So we give customers a choice of deployment…Our whole strategy is to make Teradata ubiquitous so people can consume it on whatever platform they want to use it.”
After Teradata goes live AWS, at some point you can expect to see the Aster database—which is more heavily used for big data discovery—running on the AWS cloud, Twogood says. And eventually Teradata will embrace other public cloud providers. He’s not naming names yet, but it likely means Teradata could become available on services such as Microsoft (NASDAQ: MSFT) Azure, Google (NASDAQ: GOOG) Cloud Computing, or IBM (NYSE: IBM) Bluemix.
And who knows? After the cloud, Teradata might offer its software as a packaged software product that customers could run on their own servers—or more likely, as a virtualized application that runs within a container atop a hypervisor. But don’t hold your breath. “We’re not quite ready for that type of model,” Twogood says.