Skip to Content
Author's profile photo Witalij Rudnicki

BW path to the Data Warehouse plus the role of New Technologies

There are many threads in the latest blog “Business Warehouse” by our SDN fellow Bala Prabahar. It might not be possible to touch all of the threads, so let me try to elaborate at least on some, especially those related to my previous posts.


Definition of the Data Warehouse from 90th – “Data warehouse is a subject-oriented, integrated, time-variant and non-volatile collection of data in support of management’s decision making process” is good enough, but cannot be used as an axiom today.

a) Subject Oriented – with the introduction of Enterprise Data Warehousing there is an expectation that we have multiple subjects blended together, that would allow “single version of the truth” no matter if you are getting let’s say financial numbers from finance, HR or Supply Chain report. Unfortunately I see exaggeration into the other side: putting everything into a single “EDW” making it not a data warehouse, but data waste container.

b) Integrated – one of the key words missing in many BW implementations resulting in blaming Is your BW the Data Warehouse, anyway?. What I mean is that there are lots of installations where BW has one single source system, most often SAP ERP and most often just few modules of ERP. Then someone looks at the implementation, and asks “Why did we need BW? What is its value?” Well, if they have one source system, if they are not running complex time-to-time comparisons, and if the load generated by reporting would not have sufficient impact at ERP, then the answer might be “Right, you did not need BW at all, and BW does not deliver you a value”.

c) Non-volatile – from today’s perspective even with Petabyte sized systems it does not make sense to store all the history with all details. Therefore it is acceptable to have volatile collection of data in the DW, where data older then age threshold is aggregated, archived or simply deleted.


I completely agree with Bala that current “innovation buzz” is a smoke and mirror, behind which we often do not see the real matter. I am very glad that BO+BW integration is getting real shape with coming releases of BO Enterprise BI 4.0 and BW 7.3 and that after 2 years of marketing pitch it is moving from the area of “new technologies” into “daily reality”. But now we got new subjects in the form of inMemory and Sybase.

a) It just happens that “new technologies” are more appealing than “daily reality”. In 2008 at NetWeaver BW/Portal conference I presented the session on performance of BW reporting. The section of my presentation discussed the impact of infrastructure on BW reporting, what Bala brings in his post. The feedback I got after my sessions is that it was “too technical”. It was just boring to most of the participants.

b1) It does not change the fact that some of the innovations are delivering breakthrough change. One example – memristor – is blowing my mind. Another example – BW Accelerator. I do not know who first looked at TREX (indexing of unstructured data) and saw how this can be applied to indexing of structured BW data, by calling table records “documents”, and table columns “document attributes”. But this person deserves a true innovation and out-of-the-box thinking award!

b2) The result of the aforementioned innovation is what was called HPA, then BIA, is now called BWA, and will be called HANA soon. The fact is it does deliver 10-200 faster navigation steps in BEx queries, and if someone is telling me that he can get the same improvement just by using conventional methods – I doubt.

c) Chasing innovation is becoming everything, and is becoming … annoying (please have a look as well at #2 in my Outlook of 2010: Speculations for SAP BI Professionals). With all due huge respect for SAP CTO Vishal Sikka calling every single thing developed or to be developed “an innovation” does not change my perception of SAP being more innovative, but changing my perception of the word “innovation” like not bringing something breakthrough, but as like something just being new. It is like every day of my life is innovative now, because it is a new day when I wake up, if you understand what I mean.

d) “Everything New Is Actually Well-Forgotten Old” is very often in my mind when I listen to Vital View: SAP Announcement of Analytics Solutions.

PS. Let’s discuss this during TechEd

I welcome those of you attending TechEd USA this year to join me during discussions on the topics presented above. In my role of the speaker I am very much committed to listening as well!

ASUG Educational Sessions: “Look at new EDW features of BW”

Networking sessions:

Assigned Tags

      You must be Logged on to comment or reply to a post.
      Author's profile photo Tammy Powlas
      Tammy Powlas
      I'm sorry your 2008 BW/Portals session was ranked the way it was; sounds like the audience really needed a combination of BW/Basis members.  BW performance tuning doesn't get enough attention.

      See you at TechED!

      Author's profile photo Bala Prabahar
      Bala Prabahar
      "BW performance tuning doesn't get enough attention" - Correct. Wondering why? Probable answers:
      1) BW just works fine first time. It is "self-correcting". It doesn't require any manual efforts.
      2) BW development team optimizes the code and handles the performance issues programmatically.
      3) BW development team is not interested in dealing with Basis/Infrastructure team so they try to write very simple reports.
      4) BW development team is not interested in opening a can of worms(in terms of design reviews, identifying design flaws  etc) by raising performance issues.
      5) BW is not used the way it should be used so the majority of customers don't experience any performance issues.
      6) Performance issues are too technical so a majority of people don't want to discuss about them.
      If answer is (6), then I can understand why a lot of people talk about "in-memory" databases. They probably are looking for a short answer for performance problems. They're not interested in all details, design reviews, business requirements, SAN, i/o throughput, stress testing, memory requirements, OLAP versus DB time, Network Bandwidth, DB tuning, DB or OS or SAN patching, Star Joins, index degeneration, Aggregates, bitmap indexes versus b-tree indexes, BW caching, Portal caching, ABAP performance versus Java performance, Disk space utilization, BW Compression, DB compression, table/index partitioning, global/local indexes, housekeeping jobs etc. For them, "in-memory" databases would take care of everything related to performance automagically.
      However if answer is not (5), then I would wonder why "in-memory db" and HPA/BIA/BWA/HANA concept is so popular.
      Author's profile photo Former Member
      Former Member

      I agree to (2) and (6). BW-Basis teams need a different set of technical grooming than pure ERP-Basis practitioners. Which essentially leads to (6), since if technical details are involved and no matter of what complexity, right understanding and experience helps and solves many of your performance issues even without application of BIA. Ofcourse, performance cant be matched up with that with BIA implemented.

      @Vitaliy, great piece once again!!!

      Kunal Gandhi

      Author's profile photo Former Member
      Former Member
      correction, i agree to (3) & (6)
      Author's profile photo Bala Prabahar
      Bala Prabahar
      Thanks for writing this blog. As usual, you've provided a lot of good information.
      I completely agree that BWA's performance can't be achieved using conventional methods. However I don't believe BWA is suitable for all scenarios. As long as BWA is implemented carefully, yes, the ROI would be great.
      I still haven't registered to attend TechEd. If I attend, I'm looking forward to joining you during discussions.

      Thanks again.

      Author's profile photo Witalij Rudnicki
      Witalij Rudnicki
      Blog Post Author
      Bala, with BWA the topics that you brought - memory, CPU utilization, storage and i/o, networking etc - are becoming even more crucial. "It rocks when it works" as one of my customers used to say. And it works most most of the time, but screw one of those - and the whole performance of BWA is going down. I do not expect HANA to be simple, expect it to get even more complex than BWA because besides column-oriented storage it will have row-oriented as well. Business suppose to benefit big time from new technologies, but the sooner we realize growing demand for *IT* technical mastery - the better.
      Author's profile photo Former Member
      Former Member
      Thanks for writing this blog!  You really got me going this morning to state some of the obvious that is seemingly lost with this new technology.  I'm one of the few that enjoyed your presentation @ teched and on a regular basis look under the hood at things no one else wants to touch or understand about BW. What ever happened to garbage in/garbage out?  With all these new tools/in-memory technology, people think that sound data modeling practices, conformed dimensions, subject area, single source type practices are out the window now.  Why spend good money and time on these when you can just throw memory at it and you get an answer? No matter what technology you use, your data still has to be clean, scrubbed and modeled appropriately.  Putting a tool on top and re-labeling columns on a report do not make the underlying data values harmonize.  In a scenario where you only have 1 BW system and you are dealing with master data from a rash of source systems, i don't think there is a replacement for "old school" data harmonization. The devil is always in the data and the detailed nature of it.