Insights

BW Accelerate now, or wait for H.A.N.A-ther tool?

08 Nov 2010 Business Intelligence (BI), HANA, In-Memory, Consumer Business

With all the hype around SAP’s latest in-memory innovation H.A.N.A, many SAP BW customers are wondering whether they should abandon plans to implement BW Accelerator in favour of waiting for H.A.N.A.  Hopefully this will provide some insights on the migration path and what is known about BW & H.A.N.A. so far.

What version?

The diagram below shows the version compatibility between SAP BW and BWA / H.A.N.A as it is today, as well as which major features are being added in each version:

It’s worth highlighting a few points here:

• Unfortunately there are no “in-memory” options available for SAP BW versions prior to 7.0.  Upgrading from 3.x to 7.3 is a two step process: 1 – upgrade to 7.0 & convert authorisation concept 2 – upgrade to 7.3. As of SAP BW 7.3 the 3.x authorisation concept will no longer be available.  This means that all authorisations have to be converted to the 7.0 concept first before you can upgrade to 7.3.
• The next version of BWA i.e. 7.2, is not compatible with BW7.0/1 so if you plan on exploiting some of the new features in this version, you will need to upgrade your SAP BW system.
• Moving to SAP BW 7.3 will also position you to exploit H.A.N.A once it becomes available for SAP BW.

Is my BWA investment safe?

Indications from SAP at the recent TechEd conference were that current and planned investments in BW Accelerator would be protected.  Hardware used for BWA servers could be redeployed as H.A.N.A. servers and customers implementing H.A.N.A. would receive credit for previous BWA licences bought.  That’s the general principle, but I recommend that customers confirm the details with their hardware vendors for BWA servers and SAP account managers on the licence front before they proceed with BWA, as there may be some exceptions to this.

One should also consider that the specs for the H.A.N.A. server you will need may be higher than for the BWA server due to the additional features supported and consequently larger amounts of data loaded into the server.

I’m on SAP BW3.x

If you’re planning to implement an in-memory component to your SAP BW architecture I would suggest that you look to upgrade to BW7.1 as soon as possible and consider BWA 7.0.  Moving to 7.0/1 is unavoidable and as stated above your investment in BWA 7.0 should be protected for future releases.   This will also give you time to deal with the authorisation conversion while you wait for 7.3 to move to GA (general availability).

I’m on SAP BW7.0/1 without BWA

If you haven’t done any work on converting your authorisations to the new concept I would suggest considering a start on this.  The decision on whether to implement BWA 7.0 or wait for further releases would depend on how urgent the need for performance improvement in your SAP BW system is.  Related to this is how much value indexing only InfoCubes would delivery to your organisation.  If you are doing lots of reporting on InfoSets or DSO’s then BWA 7.0 might not make enough of a difference.

I’m on SAP BW7.0/1 with BWA7.0

To move forward in this scenario means an upgrade to 7.3. The question remaining then is whether to move to BWA 7.2 or wait for H.A.N.A.  In some ways it’s almost too early to make any credible recommendations here as neither of these versions are generally available yet and the migration effort and impacts between them are not well understood. 
However, my gut-feeling at this stage is that there would be very little reason why not to go for BWA7.2 as soon as it is available. For me the compelling reasons would be the significant additional features available in BWA 7.2 and the uncertainty of what BW on H.A.N.A. would really look like when it becomes available.

Summary

H.A.N.A. for SAP BW is still some way from being generally available and in the mean time there are significant benefits that can be exploited by implementing BWA.  Add to this the fact that your investment in BWA is likely to be secured for future H.A.N.A. implementations and it seems like a no-brainer.