Sybase has unveiled a new version of the Sybase IQ high performance column-based analytics database, due to be generally available by the end of November. "This really is an extension into big data - and big data is characterized by a lot of things - but we see the trends in the market around MapReduce and Hadoop in database analytics and we have added those capabilities into IQ 15. 4," Dan Lahl, director of product marketing at Sybase, tells 5 Minute Briefing. With the new release of IQ, he notes, Sybase IQ provides customers a "have it your way" approach.
Sybase IQ 15.4 includes a native MapReduce API, Predictive Model Markup Language (PMML) support, integration with Hadoop, and an expanded library of statistical and data mining algorithms that leverage Sybase IQ's PlexQ massively parallel processing (MPP) technology. According to Sybase, the new APIs allow developers to quickly and safely implement in-database algorithms achieving greater than 10x performance acceleration over existing approaches.
The native MapReduce API enables customers to do MapReduce directly in Sybase IQ if that is what they choose to do, says Lahl. "But also integrate with Hadoop as well so you can federate a query between IQ and Hadoop. One could say I need some information from and that would actually kick off a Hadoop job as well as kick of and IQ job. And then, if you wanted to bring a result set from Hadoop into IQ, and do data federation, you can do that as well."
In addition, the ability to execute models in the industry-standard PMML format in-database within Sybase IQ allows data scientists to build predictive models in industry-standard analytics tools and then automate their execution in Sybase IQ. The PMML capability combined with existing capabilities for text and multimedia analytics is intended to provide enterprises with a range of techniques for analyzing big data.
Sybase IQ 15.4 also provides:
- A new Java API and new extensions to the C++ APIs allowing implementation of high performance proprietary or certified ISV algorithms in-database that run 10x faster. Additionally, the new in-database analytics simulation environment eases development and testing.
- Better compression of text data than commonly found in big data scenarios.
- Faster bulk loading of large data sets through ODBC and JDBC interfaces.
As far as adoption of big data technologies, Lahl says he sees a west-to-east geographical pattern among the Sybase customer base, with west coast, high technology-type companies already actively using Hadoop. These companies are finding it good for storing large amounts of unstructured data, web log information and that type of information, "but now they are asking, How do we integrate that into the parts of the analytics we are doing?"
On the east coast, says Lahl, such as customers on Wall St. that have enterprise licenses of IQ are starting to do some Hadoop work although he has not seen much of it yet, and they are asking how they will integrate Hadoop into their other analytics initiatives, and want to know what the Sybase IQ strategy is. With the new release of IQ, he notes, they can take advantage of a "have it your way" approach.
In addition, notes Lahl, IQ itself can store unstructured data. "That is why we think the MapReduce piece is going to be good for us because today the usage pattern we see is it is more on unstructured data than structured, so if you want to use MapReduce to do a combination of structured and unstructured, you can use IQ to do that. It is really ‘have it your way.' ''
Sybase IQ 15.4 also introduces a new free Express Edition, enabling IT departments, developers and application vendors to download and start developing with Sybase IQ.
Complete information is provided at www.sybase.com/iq.