Which tool do I use?

I monitor a few BI web sites and I talk with many IT folks working in DW and BI.  In spite of how many times I see and hear it, I am still astonished when people discuss which is the best BI tool.

Many years ago I heard a smart person say the whole world looks like a nail when the only tool you have is a hammer.  It seems that rule is as constant as the sunrise.

Comparing SQL Server, Cognos, SAS and Qliktech is like comparing hammers, saws, drills and wrenches.   These are all very different products with different strengths and weaknesses.  Each of them may be the best tool for something but none is best for everything in BI any more than one of the builders’ tools is the best for building a house.

My interest is in databases and I am frequently told something like “I can do that with (SAS, Business Objects, Pentaho,…) when I describe a feature of our database.

Yes, perhaps, but I guess you could use a hammer to smash the end off a board if you didn’t have a saw.  Does that make sense?  Of course not, but many people do not see the obvious parallel.

For example, the Illuminate database indexes everything, always.  This makes it an excellent foundation for analytics and information discovery.    Because of the indexing, any column of any table can be either a dimension or a fact depending on the nature of the business question.  Any column or columns of any table or tables can be used as selection criteria and there will never be a full table scan.

Can you duplicate these features with SQL Server?  Theoretically yes, but in the real world, it can never be done.  This would require building as many star schemas and indices as there are columns in the database.   You would need more disk than can be attached to a computer and most of the stars would never be used.  Unfortunately, you can’t predict which of them would never be used.  With SQL Server you can resolve some of the business issues but never all of them.
Can you duplicate these features with an OLAP product?  Even theoretically this is not possible.  How about with SAS or Business Objects or XYZ? No.  But can it be done with the right suite of products?

Again, no. The only product in a complete suite that could attempt to provide the indexing power is the database and the relational databases can not do it.  Front end systems and mathematical algorithms can not even address the issue.  There is no product or combination of products that can duplicate the benefits of the indexing function.

The real statement that people are making is that they know how to work well with a given tool and they would rather focus on the things they can do with that tool than solve business problems.  As a result, the query from hell seems to have a long and healthy future.