BIScorecard just posted an evaluation of QlikTech’s QlikView, and I confess, this review has befuddled me more than others.
The challenge with new and emerging technologies is in trying to figure out where they fit and whether or not they really are that different. So I find myself thinking about cars and bikes and QlikTech.
When cars first came on the scene, bike enthusiasts were disgusted with those smoke- spewing machines that suddenly stopped working when cars ran out of gas (interesting that bikes are making a comeback in some cities). Sometimes innovations call for new evaluation criteria – with bikes and cars, features like pedals and gears simply don’t translate against miles per gallon. So does “in-memory” BI make criteria like SQL-generation less relevant? Why do you need a data warehouse at all if you can load a full Terabyte of data in-memory?
In talking with QlikTech customers, all enthusiastically declare they have waited years for this level of empowerment from either their central IT group or from other BI tools they’ve tried. Some of their comments are similar to those of Excel users who later (and still) suffer the consequences of spreadmart chaos. Other comments are reminiscent of early MOLAP (Oracle Hyperion Essbase, IBM Cognos PowerPlay, Microsoft Analysis Services) tool users who could build interactive data marts without IT’s involvement.
So is QlikView really all that different or is it just a re-incarnation of personal and departmental analysis tools? I think the answer lies somewhere in the middle. Their best “feature” is the rapid implementation time, without as much data chaos as spreadsheets and with more flexibility than traditional MOLAP tools. Sometimes speed-to-insight trumps a perfectly architected data warehouse and broad BI solution.
By the way, QlikView is not the only product that does not neatly compare to a single module within a BI platform. Tools like TIBCO Spotfire and Business Objects’ recently released Polestar face similar challenges with product positioning. Even advanced visualization tools such as Tableau, Advizor Solutions, and Corda sometimes struggle with articulating where they fit in the BI spectrum.
Regards,
Cindi Howson
Hello,
In a recent discussion on TDWI's LinkedIn group it was pointed out that this "just load, then query" without any modeling or data preparation in between, is a bold claim by some BI tool vendors. Some information might be easy to understand and load directly from the source OLTP system, but in many cases metrics have to be prepared, cleansed and modeled, and some times consolidated from several sources to produce something meaningful for the business.
I've tried QlikView's personal edition and it's an interesting approach very visually apealing tool but, it still requires technical expertise for the regular end user. As you say, "the answer lies somewhere in the middle". It might be very powerful for departamental or even corporate use, but best and fast results will happen only very reliable and well modeled data.
Posted by: Ricardo Sada | December 23, 2009 at 01:09 PM