Take a powerful database and put a thin business intelligence application on top; that's the combination behind a couple of announcements last week including SAP's HANA appliance and an Aster Data/Tableau Software partnership.

These announcements are next steps in the trends toward in-memory computing and in-database analytics -- trends that could redefine the business intelligence suite.

Taken together, these trends are taking the query part out of BI's traditional query-and-analysis role and putting the emphasis strictly on analysis. The latter is where BI meets the business, so all the better if it's fast, flexible and intuitive software rather than something geared to IT.

The headline promise behind SAP High-Performance Analytic Appliance (HANA), released last week, is the ability to tap into real-time transactional information. But that's not the only argument for the technology. Raw processing power and in-memory access to data promise to give ordinary business users the ability to ask questions that weren't possible -- or at least practical -- before.

In-memory BI products like QlikTech's QlikView, TIBCO Spotfire, IBM Cognos TM1 and Advizor Solutions have already taught us that business users will explore and find answers on their own if they can get at the data through intuitive interfaces and with reasonable query response times.

It's also known that users won't explore if it means they have to submit requests to IT for ETL jobs or new cubes or even if they have to go through interim aggregations, calculations or minutes- or hour-long queries.

So we can all agree that in-memory is a good thing, but SAP and others are giving us options as to where to put that power. For now, HANA is an add-on to an SAP Business Warehouse or a BusinessObjects environment, but the next step will see it become the warehouse itself.

In fact, SAP ultimately sees the technology becoming the infrastructure for applications as well, so you won't even need a separate warehouse. This last step is years down the road and has yet to be proven, but let's assume that next-generation technology will get us there.

The bottom line is that query processing power and, increasingly, in-memory capabilities are showing up in analytic appliances. Kognitio and others have already implemented massive random access memory. And as announced in June, SAS will likely be next to deliver an in-memory appliance -- in this case supporting industry-specific applications.

And then there's Microsoft's planned Denali release with Crescent, the latter a code name for a Web-based, ad-hoc data visualization and presentation environment designed to improve data exploration and analysis. No doubt Crescent will give business users in-memory power with a bit more hand holding and guidance than available in the power-user-oriented PowerPivot add-in for Excel.