Original URL: http://www.theregister.co.uk/2013/02/28/paraccel_analytic_database_gets_sql_overlay/

ParAccel plugs SQL queries into its analytics engine

Thar be money in them old SQL hills, boys!

By Jack Clark

Posted in Applications, 28th February 2013 17:17 GMT

Strata 2013 Amazon-blessed database company ParAccel has layered an analysis suite for SQL users on top of its technology.

The extra analytic capabilities – social-media parsing text features, sessionization, JSON parsing, and pattern matching – combine to make the complex system a little bit more attractive to old SQL pros, ParAccel announced on Wednesday.

"A common challenge for ingesting web data is understanding the characteristics of sessions," Walt Maguire, ParAccel's analytics director, told The Register.

Trendy tools like Hadoop can be clunky to use when you need to generate new analytical reports, Maguire said, while existing tools for getting a broad swathe of insight into how people are behaving on your website entails paying a big wodge of cash for a service like Omniture.

ParAccel hopes its new analytical suite will let the database sit somewhere in between the two, giving it a little bit of the flexibility for on-the-fly reports of a typical online metric suite, and some of the computational heft of a platform such as Hadoop.

"In Hadoop it's a brittle, time-consuming process to flex [data] – a lot of people pay Omniture for that," Maguire said.

Some of ParAccel's features, such as tokenization, can help companies analyse social data, such as when someone tweets hashtags that are highly relevant for monitoring the spread of a certain keyword but are (mostly) useless if running large-scale sentiment analysis.

Another feature includes stemming and spell correction, so the system can group various misspelled messages (ones containing "your" along with other keywords can be linked to ones contained "yor" or "yr" or other common abbreviations and howlers) together for further analysis.

ParAccel's features are part of a growing trend in the database and "big data" industry to embed some SQL-like querying into its distributed unstructured data products. The database, cognizant of this, already comes with node-to-node Hadoop connectors that let it transfer data and processes in and out of Hadoop in both directions.

"There's a huge race on right now for SQL access to Hadoop data," ParAccel's vice president of solutions John Santaferraro told The Register. "When we look at some of what's being offered to the marketplace, they're taking a little sliver of SQL capabilities," he said.

The functions will be available within the first half of 2013. The company is still deciding on whether to charge extra for the features or bundle them with the existing ParAccel database. ®