Intelligent Enterprise

The headlines emerging from this week's Teradata Partners Conference in Washington D.C. will no doubt dwell on the vendor's Enterprise Analytics Cloud initiative and its planned release of an Extreme Performance Appliance based on state disk (SSD) technology. Both announcements deserve close attention given the flexibility promised by cloud approaches and the incredible speeds promised by SSDs. But for all the promise, Teradata is sticking to a fairly conservative strategy and a surprisingly understated public announcement at this week's big event.

Teradata's four-part cloud initiative combines two offerings for private-cloud deployments behind corporate firewalls and two aimed at public-cloud deployments. As with most cloud offerings, the goal is to offer the flexibility and business-user-friendly advantages of self-service provisioning -- albeit mostly within the bounds of conventional, on-premise Teradata deployments. In fact, most of the functionality behind the cloud initiative is already offered though Teradata's Active Workload Management capabilities. What is new is an Elastic Mart Builder provisioning tool that provides an easy, graphical user interface for requesting, loading and deploying temporary marts for test, development and sandbox-style experimentation. Teradata security and access controls apply, but with self-service provisioning, IT no longer stands in the way when users want to access their data, experiment and, hopefully, innovate.

The self-provisioning functionality is available immediately through a Teradata Viewpoint portlet that can be downloaded for free through the Teradata Web site. The second private-cloud option is a Teradata Express client, now in beta, for the VMWare player. This virtualization option will be limited to four nodes and 1 terabyte of data.

The first of the two public-cloud options announced today is Teradata Express for Amazon. The EC2-based service, set to bow early next year, offers the public-cloud benefits of paying only for the processing power that you use and turning it on and off at will. However, it's also limited to one node and up to 1 terabyte of data -- a tiny cloud to be sure. Aster, Kognitio and Vertica have preexisting EC2-based deployment options with few such restrictions. Teradata executives say they intend to continue developing public-cloud options. But for now, larger and more capable options will have to wait.

"Any type of shared resource has contention on CPUs, memory and I/O [input/output]," said Randy Lea, vice president of product and services marketing at Teradata. "Right now the cloud providers don't have the I/O-service-level delivery that you would expect in a data warehouse production system."

The second public-cloud offering is a software-as-a-service partnership with SilkRoute Global, which will offer a Demand Chain Management forecasting application for retailers. The service will let retailers upload point-of-sale information and other data up to Silk Route's Teradata-based service to help retailers anticipate item-level demand by product, store and other granular dimensions.