Teradata data warehouse joins SSD technology to HDDs, virtual storage

At Teradata Universe in Barcelona, the company announced the Active EDW 6680, a data warehouse system that mixes SSD technology, hard disks and storage virtualization software.

BARCELONA, Spain – The world of analytics never stands still – users always want to analyse more data faster than ever before. Teradata has for some time had two technologies that cried out to be plugged together because they promise just that capability. One is solid-state drive (SSD) technology; the other its Teradata Virtual Storage (TVS) software.

The question was never about whether Teradata would combine these technologies but when. The answer came April 11. At its Teradata Universe conference here, the company announced the Active Enterprise Data Warehouse 6680, an EDW system that includes both SSDs and conventional hard disk drives (HDDs) and automatically migrates data between the two storage technologies based on usage levels.

SSDs are much, much faster than rotating hard disks. SSD technology in laptops is a given – many people are finding the faster boot times and increased shock resistance well worth the higher price. Also a given is that most analytical systems are limited by disk I/O; as a result, moving the data being analysed from rotating media to something faster is an obvious solution. In 2009, Teradata took a first step down that road, expanding its line of data warehousing appliances by announcing the Extreme Performance Appliance, which is based entirely on SSD technology.

Of course, Teradata isn’t the only vendor to have worked out that faster media equals faster analysis. There are, for example, companies that offer entirely in-memory analytics, which is even faster. An important limiting factor here is cost – the faster the media, the more expensive it is. If you have only a few gigabytes of data, the best answer is to put it in RAM. If you have petabytes, RAM is unlikely to be a cost-effective solution.

However, analysis of analytical systems reveals that the data they contain is not accessed evenly; in practice, the majority of queries are run against a small subset of the total data. The catch is that this subset changes over time. When data is initially loaded, it typically is “hot” in analytical terms. With the passage of time, it cools – in other words, it is accessed less and less frequently.

Also since 2009, Teradata has offered TVS, a storage virtualization tool that automatically puts the hottest data onto the fastest media in a system and then migrates information to slower storage devices as usage declines.

Enter the Active EDW 6680, which uses TVS to move data between SSDs and regular disk drives and can scale from 7 TB to 32 PB of storage capacity. In addition, there is a new Active EDW 6650; it comes standard with HDDs only, but Teradata said that SSD technology and TVS software can be added at a later date.

Teradata claimed that the new 6680 delivers up to four times the data warehousing performance of HDD-only systems with the same capacity. The 6680 also provides up to 75% lower energy costs for the same performance levels over previous appliance models, as well as a 75% smaller data centre footprint, the company said. The figures for the 6650, which supports up to 92 PB of HDD storage, are 25% lower energy costs and a 25% smaller data centre footprint.

Read more on Data warehousing

Data Center
Data Management