Teradata gets its cloud and NoSQL on as big data pressures mount

Business analytics pioneer Teradata (s tdc) continues to keep up with the Joneses in the big data world, announcing on Monday a handful of new products and features — a hosted cloud version of its flagship database (which Netflix (s nflx) is already using) and support for JSON documents among them.

The Teradata Cloud is just what it sounds like — the popular Teradata data warehouse delivered as cloud service instead of as a physical appliance. It will have all the functionality of the traditional appliance, but is charged on a a subscription basis and allows for fast provisioning of new capacity, Teradata Labs President Scott Gnau said during a recent interview. The company expects it to be “TCO-neutral” with the physical appliance in terms of price, he added, and will add Aster Data and Hadoop as a service in the first half of 2014.

Gnau didn’t seem too sold on the idea of data warehousing and analytics delivered as a service (“I don’t think it’s a mainstream requirement in the data warehouse industry today,” he said, citing possibly overblown fears over security and privacy), but acknowledged there is a market already shaping up. Mostly, companies such as Netflix with cloud-first mandates should be interested, as should mid-market companies such as Teradata Cloud customer BevMo. In theory, the cloud version will let them grow into their data warehouse needs, save them the hassle of managing the physical appliance and also make it easier to test out new workloads.

teradata cloud

Source: Teradata

If Teradata is a few years late in really getting into the cloud game, though, it’s trying to get ahead of the curve with regard to the internet of things. Gnau called JSON — a popular data format to house in NoSQL databases — “the language of the internet of things” and said Teradata users will be able to store JSON files without predefined schema. Analysts will be able to query the files as they see fit and when they see fit using SQL or certain NoSQL languages (startup MemSQL also now supports JSON files), or integrate the JSON-stored data with traditional relational data.

Gnau thinks Teradata customers, which tend to be rather large businesses, will be especially eager to use the new functionality for collecting data from sensor networks. He presented a scenario where a company might track down products affected by a recall by using time-series queries to pinpoint the suspect units, and then joining those results with relational data to figure out where it sent those units.

There has been quite a bit of chatter lately about whether some of Teradata’s revenue woes are the result of open-source technologies like Hadoop and NoSQL undercutting the value proposition of its high-cost data warehousing appliances and proprietary software. I happen to think there’s truth to this theory (others vehemently disagree), which helps explain why Teradata is putting so much effort into supporting Hadoop, JSON, graph analysis and other new workloads, and pushing out ever-less-expensive product lines for bulk storage.

But I also think Teradata’s efforts to support new types of data types and workloads is a strong indicator that it sees that challenges ahead for its business and is trying its best to address them. What is yet to be seen is how big an effect the open-source-from-the-start technologies like Hadoop will actually have on large vendors’ bottom lines. Even if companies like Teradata are successful in attracting more customers and more data, they might have to get used to less-profitable deals in the age of open source.