Traditional BI technology loads data onto disk as modeled tables and multidimensional cubes, queries are then made against the tables and cubes on disk. Limitations of disk based (RDBMS or OLAP) techniques include performance limitations requiring intermediate aggregation tables, low flexibility to adapt to changing business needs, limited scope of analysis and long implementation cycles.
In-memory technology removes these steps, as data is loaded into RAM and queried in the application or database itself. This greatly increases query speed and lessens the amount of data modeling needed. Memory prices keep dropping which makes it economically viable to increase capacity for in memory processing. Faster performance on larger data sets with less data management seems like a win-win situation for the organizations.
While data warehouses guarantee integrity and provide a stable server environment for managing data, in-memory can make information accessible at the time it is needed and available to anyone who requires it.
“In-memory technology is not in and of itself a driver of BI growth, despite the massive hype of many vendors. To drive adoption, in-memory must be coupled with consumer-oriented BI tools, a combination that has been at the heart of data discovery tool success. As they have been to date, in-memory capabilities will continue to be an enabling technology. They will expand BI to a broader range of users, as more and more BI vendors incorporate it into their portfolios to deliver Google-like responses in exploring vast amounts of increasingly diverse data types via intuitive, yet sophisticated and mobile BI tools and applications.“, Gartner Jan 2011.
There are few articles I would recommend reading including Top 10 technical requirements your in-memory analytics vendor by James Mandrid, What to look for from you In-Memory BI platform by Qliktech and finally my favorite from Boris Evelson, at Forrester on Not All In-Memory Analytics Tools are created equal. Another interesting article you might enjoy is by Elad Esraeli where he talks about how In-Memory is not the future, It’s the past.
Qlikview, Tibco Spotfire, IBM Cognos TM1 (formerly Applix TM1), SAP BusinessObjects, MicroStrategy, Microsoft, Tableau are some of the in-memory vendors in the market. Personally I think in-memory is not the best option for scalable, multi-user BI apps. 64 bit computing on column based technologies can provide another alternative to hefty OLAP projects.
Why there is all the hype on in-memory BI?, Why it has become so fashionable?, How new and innovative it is?, Will it kill off disk based BI and the is th next breakthrough and What is the next breakthrough?. A report on What in-memory BI ‘revolution’? by Business Application Research Center (BARC) would answer most of your questions along these lines.