Teradata has got a new purpose-built appliance for SAS high-performance analytics that uses an in-memory approach for hyper-fast results.
In other words, it distributes complex analytics in parallel across a vast pool of memory looking for patterns in large volumes of data.
It reportedly whittled what would normally have been a 167-hour project in financial risk analysis at some Wall Street bank or another down to 84 seconds.
Teradata claims other customers can expect as much and expects it to "kick competitive butt." It claims IBM, Oracle and SAP, which have their own in-memory systems, "lack the foundational analytics."
It's good for data exploration, model development and model deployment. Besides compressing analytic modeling life cycles from weeks to minutes Teradata says it will dramatically increase yield from current analytic modeling processes, even solve pesky business problems previously unanswerable.
It's thinking telcos, retail and financial services.
The integrated Teradata 700 appliance is called logically enough SAS High-Performance Analytics for Teradata since this particular Teradata 700 model only works with SAS' software. It also integrates with Teradata's so-called Analytical Ecosystem so it can be synchronized with other systems in a user's infrastructure.
The Xeon widgetry, based on SuSE Linux 10 and the Teradata Database 13.10, comes in small, medium and large configurations. The entry-level small offers 192 processing cores, 1,536GB of RAM and 20.4TB of storage. The medium has 288 cores, 2,304GB of Ram and 30.6TB of storage and the large has got 384 cores, 3,072GB of RAM and 40.8TB of storage.