Friday, April 19, 2013

Working with SAP BW in Tableau

On June 25th 2012, Tableau announced that they achieved SAP Certification for Integration with SAP Netweaver (BW).  SAP users all over the world exclaimed with excitement with the news that they would finally be able to tap into the data collected in SAP BW.   Just a month earlier, Dr. Vishal Sikka announced SAP Visual Intelligence at SAPPHIRENOW in Orlando, Fl.  While the product has grown since its initial offering, it is still only able to connect to CSV, Free hand SQL, HANA, MS Excel and BO Universe.  So why wouldn't SAP want to be able to connect to data in SAP BW?   Well, given my experience working with it, I think I can guess; it's a pain.   Here's why.  SAP BW sits on top of operational systems. (CRM, ERP, R/3, FICO, etc.)

Image courtesy of SAP
All of this data coming in becomes too cumbersome to work with, so Cubes are used to be able to find answers from the data faster.  Problem is, data is endless, and the size of these cubes only gets bigger and bigger.  Eventually you lose the benefit of the cubes because the business needs a faster, more flexible way of working with the data and SAP BW simply can't provide that.  Think I'm mistaken?  Well I always encourage an open dialog, and I'll be the first to admit I don't know everything.  However, consider this, if SAP BW was an efficient EDW to query off of, why would SAP offer a SAP BW Accelerator product with promises of improving query response times by up to 200x?  By the way, if you want that, that will cost you roughly 6 figures per 16gb blade.

Can Tableau connect to SAP BW?  Yes.  Is it fantastic for visualizing the data contained in that environment?   Yes.  Is Tableau fast?  Yes.  Is SAP BW fast?  It depends.  Is SAP BW flexible for analytics in Tableau?   Not really, though this once again has to do with the data structure itself, cubes were never meant to be 'flexible.'

While we do support creating calculated members using MDX, yet another drawback with BW is that SAP MDX is different.  Different you say?  Well SQL is SQL because its a Standard. Shouldn't Multi-Dimensional eXpression language or MDX be standard?   Not in SAP world. Say goodbye to familiar functions like CDATE, or any MDX functions involving strings.  The bottom line, set expectations appropriately, know the constraints of the technology you are working within.  Tableau is a great way of being able to understand the data that is in SAP BW like never before, but that doesn't automagically make working with SAP BW data for analysis a good experience.

So, lets bring it back to the beginning.  SAP announced a tool that lets you visualize data coming from HANA.  Well, guess what, Tableau connects natively to that also, among several other databases.  What is to be expected?  Well, HANA is an in-memory relational database, and while I haven't worked with it personally, I am definitely excited to get my hands on it with Tableau.