On February 22nd Aster Data makes available version 4.5 of their massively-parallel data warehousing and advanced analytics software. Aster Data specialises in complex analytic processing of high data volumes that require near real-time analysis rather than the regular batch data processing and reporting that comprises much of the workload of traditional corporate data warehouses. For example Full Tilt Poker uses Aster Data's technology for fraud detection. Previously this processing was so complex that they could only run it on a sample of the games in operation on their web site; it took 90 minutes to finish and was run weekly. The company are now able to finish the fraud analysis on the complete set of poker games within 90 seconds, and run this analysis every 15 minutes, dramatically reducing their exposure to fraud. Other examples of this kind of complex analysis are retail basket analysis, assessment of stock market trading positions, and click steam analysis. In one case, a telco uses Aster Data to carry out detailed analysis of call patterns in order to discover the usage patterns of various demographic groups (such as teenage girls), and then try to sell these customers additional services that will be particularly relevant to them.

The key to Aster Data's ability to carry out this workload has been their early development of the MapReduce framework (a model for processing large datasets on a distributed platform which was popularized by Google) within their database engine, and tightly coupling it with standard SQL. Actually coding for this type of query is usually the domain of specialists, and in version 4.5 Aster Data has made it easier to invoke such processing by adding an integrated development environment (IDE) for MapReduce. This provides common pre-built functions for analytics, like moving average calculation and time-series analysis, and a series of wizards to allow these functions to be easily incorporated into the database. Aster Data already supported Java, C#, and other common programming interfaces to MapReduce, allowing a wider range of programmers to be able to take advantage these capabilities than most other MPP database vendors.

Additionally, greater speed of analytic execution is provided in version 4.5 by allowing support for solid state storage, and a faster loader allowing practical load rates of up to 4 TB an hour. Another major theme of the release is improved ease of operation, provided in this case by a new management console that allows graphical displays of the queries and other other analytical processes that are running, the activity on the various servers, etc.

In a crowded data warehouse market it is important to stand out from the crowd, and simplistic marketing messages that just say "we are quicker and cheaper than the big boys" are insufficient to enable prospects to distinguish which data warehouse platforms are best suited to which kind of processing use case. By concentrating its messages around its competitive advantage in complex interactive analytics, Aster Data is sensibly trying to carve out a niche for itself, attacking a high-value segment of the market rather than taking a scattergun approach. With a much expanded number of quota-carrying sales reps now in place Aster Data must now see whether this targeted message resonates with the market.