IBM 'Smart Analytics System' Aimed at Reinventing Data Warehousing

Big Blue says the new IBM Smart Analytics System will deliver optimized combinations of hardware, software and business-problem-specific applications that will reset deployment-speed, cost and performance expectations for "analytics-ready" data warehouses. If the company can deliver -- and there are certainly questions and competitive threats -- it may shake up the piece-part integration approach that has prevailed in data warehousing and business intelligence.

The IBM Smart Analytics System is to be formally announced and available in September, and the company also announced an IBM Smart Analytic Optimizer designed specifically for IBM System Z computers. The latter is to be introduced in the fourth quarter and will act as a coprocessor supporting analytic queries and cubing against data in live, mission critical systems running on System Z computers. Thus, the thousands of government agencies, financial services, retailers and manufacturers still running these mainframes will gain an option for query and analysis capabilities without having to build a data warehouse. The Optimizer will handle all query and analysis processing so that work won't affect the performance of the host mainframe.

Questions Abound

IBM executives stressed that the Smart Analytics System will be purchased and shipped as a single product that won't be available in a-la-carte fashion.

Q: Is this simply a broader repackaging of IBM's existing InfoSphere Balanced Warehouse offerings?
A: "This system delivers an order-of-magnitude deeper integration and optimization than in the Balanced Configuration units," responded Steve Mills, IBM's senior vice president and executive of the IBM Software Group. "We've optimized the structuring of data, loading data, the queries -- it gets into all the fine-grained aspects of system-level optimization, so it's beyond a packaging approach."

Q: Are the optimizations predefined or are they customized for each customer and deployment?
A: "The answer is both," responded Arvind Krishna, IBM's vice president of enterprise information products. "We're doing the bulk of the optimization up front in manufacturing before the product ships, but if the customer wants more work done to optimize a particular application, that has to be unique to the customer."IBM executives made numerous claims about faster, optimized performance and lower cost, most of them tied to IBM's unique ability to bring together all the piece parts -- from hardware to systems management to applications. "We believe the cost of the total system is reduced by at least 50 percent [compared to the build-in-yourself approach]," Krishna said. "If you buy the storage, operating system, hardware, and warehousing and analytic software a la carte and then you integrate everything yourself, can you imagine the nightmare of version gridlock or the puzzle of optimizing it to work together more efficiently? IBM has taken care of all of that and we support it all under one contract."

The Competition

Plenty of data warehousing vendors have combined storage, servers, networking and database management systems that can be purchased and supported as a single product. But IBM is upping the ante particularly by bringing BI software and industry-specific content into the same package. Executives also made it clear that technologies from pending acquisitions, such as SPSS, and past acquisitions, such as ILog, would be deliverable through the new system. If the combinations prove to be broadly applicable and appealing, it would challenge the status quo in data warehousing. Oracle could likely muster a response by combining its Sun hardware, BI software and applications, but most other vendors, from Teradata and Microsoft to Sybase, Netezza and Greenplum would have to forge new partnerships.

The competition that IBM didn't mention yesterday was commodity hardware and open-source software, which have already been powerful agents of change in data warehousing.

"When companies see the performance they can get out of cheap hardware and open-source databases, they are putting these systems into production," says Jim Davis, senior vice president and chief marketing officer at SAS, a rival of both IBM Cognos and SPSS. "The computing power that we're seeing out of those types of environments is going to challenge the Oracles, the IBMs and, to some extent, the Microsofts of the world."Though IBM is promising better performance, a big part of the appeal seems to be targeted at executives who would favor contract simplicity and a single "throat to choke" over enterprising, but potentially riskier, in-house development, integration and innovation.