Does Qlikview need a DWH? The sage continues

Posted by Gilles on 04/06/2009 under Articles | 4 Comments to Read

In an earlier post I asked myself if Qlikview needed a datawarehouse or that Informatica needed a frontend. This discussion is going strong ever since and ever before. Qlikview sales and Qlikview fanatics claim that there is no need for a warehouse, Qlikview can do it all! I certainly do have my doubts about that! Ronald Damhof a Dutch Information Management Architect, who is much respected by myself as well by the BI Community in The Netherlands is strongly agitated by the message sent out by qlikview fanatics.

He came across this post here, (a very coloured and not to toughtful blog) which triggered him to react on this here , which eventually leaded to a bigger and little agitated reaction here. I strongly agree with Ronalds points in his reaction, but I’d like to add to the discussion that Qlikview can also act as an enabler for new Business Intelligence initiatives or as a quick problem solver.

In a company, where there is no formal BI program and excel sheets are created everywhere and send anywhere to support most decisions, Qlikview can be the start of a more formal BI program (as any tool could be). The strong points of Qlikview here are its rapid deployment, quick results and a quick ROI. So Qlikview with no datawarehouse could be a good option for a company with a low level of maturity on BI. The risk here is that Qlikview works so good, that it easily can replace excel for the BI needs and what you’ll see happening is Qlikview applications all over the place, with no allignment and a lot of departemental BI solutions.

When BI needs grow, requirements become more clear. That is where we, BI professionals, should help our customers on there way to maturity. With a good (architectural) development of the BI program, both business wise as technological. Not selling a mamoth when a mouse is needed, but designing a great mouse with the mamoth in mind.

Share and Enjoy:

Tags: architecture, data warehouse, qlikview

Gilles said,

And here is Ronald’s reaction: http://prudenza.typepad.com/dwh/2009/06/the-qlikview-chronicles-the-sage-continues-ii.html
Juan said,

Great post Gilles. In my opinion the discussion of whether QlikView can replace a warehouse or not must be put in perspective to what is actually happening right now in the market. More often than not companies that get started with QlikView have already invested in a DWH so it doesn’t make any sense to reinvent the wheel. In those cases QlikView is the perfect match for extracting the data within the warehouse an making it available to everybody in the most cost/effective way.
For companies that are just getting started with BI (rarely big companies), it is a question of volumes. How much history and detail to store and the particulars of the industry in which the company is operating are factors to consider. If the volumes are easy enough to handle then a QlikView solution without a DWH may work well. But if the complexities of keeping/storing historical data, and the requirements of data integrity (ie consolidating data from various source systems) are too difficult to handle with scripting and QVDs then a warehouse might be desirable.
All this back-end talk is independent of the specific reporting/analysis needs of the end users. We can think of a situation where some QlikView apps consume data from the DWH while others don’t need the warehouse at all, and may extract data directly from a particular source system/s. This a very flexible approach, because it frees the DWH from becoming a unnecessary monster or bottleneck to deploy simple solutions quickly.
Gilles said,

Juan,

I can agree with you for the biggest part of your reaction. I cannot agree with you on your last paragraph where you can think of a situation where some QV apps consume data from outside the warehouse. I can think of that as a temporary solution, but we all know that temporary solutions in the end will become permanent solutions. Which will result in an explosion of QV applications and then, what’s the difference with Excel.

In my opinion you should always try to make the datawarehouse the single entry point to information. The datawarehouse is the place where you can maintain sound definitions, make sure that data quality is high enough to base decisions on, data is clean, etc. etc.

I do agree with you that the dwh could become the monster preventing quick and simple solutions, but that has a lot to do with good architecture and design of your warehouse. Two good articles on architecture (and a third coming up) is available for download here: http://www.prudenza.nl/nl/library/

Cheers,
Gilles
Christian Bader said,

Hi,

QlikView is great for data analysis and a small company could start without a underlying data warehouse. You could even build your data warehouse with QVD files, but you would need to build a lot of functionality (eg. transactions, updates, etc) yourself. Common DBMS are delivering this. Another problem is, that QlikView is not easibly able to deliver data via ODBC/OLEDB/WebService, etc. for other application, which a normal DWH is typically doing.

I would suggest the following architecture. First you are storing your data in a simplified DWH (eg. large tables, no extensive precalculations, no star/snowflake schema, etc.) and then you are loading your data to QlikView (QVD and QVW). I think, that this is the most valuable approach for end users and other applications.

Regards
Christian