An Apache Hadoop appliance is under consideration for the IBM Netezza product line but the vice president of product management said that they will not develop one unless it can be a game changer.

Price performance and simplicity would be a critical factor, said Phil Francisco, vice president of product management for IBM Netezza during an interview with me today at the IBM Information on Demand conference in Las Vegas. Ease of use is another issue. Francisco said one of the things that stymies more innovation is Hadoop’s complexity. The early adopters of MapReduce and Hadoop brought their own level of expertise to the enterprise. But as you get into a much wider set of users that level of complexity is not tenable.

Netezza would differentiate around its increasing emphasis on application oriented appliances, designed for a broader set of users. A different level of risk assumption and expertise would surface with a Hadoop appliance but the approach would be similar. Netezza does not currently have a set road map for a Hadoop appliance.

This past May, EMC Greenplum unveiled a Hadoop appliance. At OpenWorld, Oracle announced Oracle Exalytics, which is supposed to be a Hadoop appliance as well. And yesterday, Karmasphere unveiled its own virtual Hadoop appliance, designed to build jobs on IBM Infosphere BigInsights. According to a Karmasphere press release, the bundle incorporates BigInsights, CentOS Linux, the Eclipse IDE and Karmasphere Studio Community.

IBM Netezza now has the ability to do some basic Hadoop processing and ETL processing of structured and unstructured information.

Francisco said more importantly for IBM is the Infosphere Streams tie into the platform which gets back to Netezza’s emphasis on simplicity and application oriented appliances.

Infosphere Streams does the more real-time stream processing on the Netezza 1000 platform. Netezza does longitudinal analysis on the data as well as provide operations with more longer term help with decision making. Netezza also provides feedback loops to do pattern analysis on the data streams. These types of deployments are being used in the financial services sector, telecommunications and smart grid types of deployments.

Netezza also has a Hadoop connector for BigInsights based on Jaql. Data can be pushed into an appliance. There can be pre-processing and pushing data into the data warehouse or they can be pulling structured data out of the data warehouse into the analytics they are running on the Hadop grid. Netezza has a similar connector with Cloudera.

Services Angle

Netezza is making deeper forays into appliance development. At IOD, IBM announced a network analytics accelerator designed for communications services providers going from 3G to 4G service. It is bundled with the Netezza 1000 analytics platform, IBM Cognos and models and tools for analysis from Ventraq, which specializes in telecommunications analytics.

The application appliance play has a number of business cases such as CRM or enterprise content management. Think Sugar CRM and Alfresco Software as examples in which the appliance serves as a way to service a heterogeneous environment with integrations into third-party infrastructures.