Head of Business Analytics & Technology
Bluefin Solutions
Is NetWeaver BW really dead and was it really killed by HANA?
02 Jun 2011
Business Intelligence (BI), HANA, In-Memory, SAP NetWeaver Platform, Business Objects, Consumer Business
There has been a lot of noise about HANA ringing the death knoll for SAP's Enterprise Data Warehouse, and most of this analysis holds no water with reality.
If we get to the facts of why organizations implement Enterprise Data Warehouses, it is usually for a number of reasons. Here are the most common:
- Pulling information from multiple assets in the enterprise into one location
- Providing a single view of the truth - i.e. of product, customer etc.
- Performing complex business transformations so information is easily digested
- Performing aggregation so information is consumed at e.g. product group, country levels
- Reducing the cost of information management by storing enterprise information in one location
SAP's NetWeaver BW Data Warehouse is a very good product that does all these things very well, but it has a few downsides. The first is agility. BW causes IT to build complex structures which IT then builds change control around, and this often makes change slow and difficult, which causes a perception of inagility. The second is performance. Very often, best design practice is not followed and performance is poor - exacerbated by a poor take-up of SAP's Business Warehouse Accelerator in-Memory product.
What does HANA fix and what does it not fix?
Basically, the current version of HANA (1.0) fixes agility and performance (in very broad terms) but it is not an Enterprise Data Warehouse so doesn't do any of the things that NetWeaver BW is currently good at. HANA 1.2 will be as good as BW as an Enterprise Data Warehouse and should perform well, but won't be any more agile than BW - because it is BW. Organisations will be still looking at ways to become more agile, though faster load times and fewer layers of Data Warehouse will help, in time.
Why Business Content is so important
One of the most important things about NetWeaver BW as a Data Warehouse for existing SAP Business Suite customers is the value of Business Content. Basically it comes with a pre-built set of Extract, Transform & Load (ETL) tools to connect NetWeaver BW to the Business Suite and get information out, pre-built into structures and reports.
By the way, most of the value is in the Business Contract Extractors - that provide the logic of how information in ERP is connected - because of the way that they handle changed information without reloading all the information from the Business Suite. There is much less value in the later data structures and reports that BW provides because they are relatively quick to build and most customers want something different.
Without Business Content, using HANA 1.0, Sybase IQ or any other database like Microsoft SQL Server for example, is expensive because it requires building out content by hand. For a point solution or datamart this may make sense, but for an Enterprise Data Warehouse, it is very expensive.
Why is Data Services the elephant in the room?
When SAP acquired Business Objects, it brought with it the product formerly known as Acta. This is an Extract, Transform & Load tool for any scenario and is quite a powerful and best-of-breed product. It can be used to extract information from SAP's Business Suite and to load it into NetWeaver BW but this was previously only useful for integration of non-SAP Data Sources into NetWeaver BW.
And a colleague pointed out to me this week that with SAP BusinessObjects Data Services 4.0, things change because you can use the Business Suite Business Content Extractors using Data Services - to extract from the Business Suite. Due to the way that Data Services works, this means you can load into any database - like Sybase IQ, Cognos or indeed HANA 1.0.
BW is dead, long live BW
Now, SAP does not really care what you build your Data Warehouse on provided that you use the Business Suite - and preferably the SAP BusinessObjects analytics tools, because they don't derive any substantial additional license revenue from BW customers. And the reality is that for existing BW customers, there is no reason to change.
However for customers who don't want to use BW because e.g. Cognos is their strategic Data Warehouse, or who don't have a Data Warehouse yet, this gives options that they simply didn't have before, without incurring massive consulting costs. And for those who do use BW, they will continue to do so.
Does it kill BW - no - but it certainly a more credible threat than HANA!
Ian 27 Jun 2011
I was told by a SAP employee that they have already been breifed on the roadmap for BW and it will be 'retired' within 7 years and will not exist after that.
So yes BW as we know it will disappear to be replaced by DS / Hana etc. There will still be a warehouse type layer but it will be a completely different beast.
Luc Vanrobays 23 Jun 2011
Well, the very good news is that HANA SPS02 that came out GA last monday comes with business content (analytical and attributes views) and you will probably have to add-up your Y fields, but the data provisioning is build-in. Up to you to create the calculation views to integrate the models and enrich the delivered content.
All of this supported with simple SQL scripts (executed at IMDB level) and own defined table types that any end-user tool will consume, even Web Dynpro ...
The other good news is about Agility, because HANA SPS02 offers Near Real Time replication with a dedicated NW 702 SLT instance allied with db triggered log tables on Business Suite.
Happy Winter (in South Hemishere) !!
Anand Gupta 14 Jun 2011
The danger is not that BW will be killed straight-away but will be chipped off slowly.
I think the game has moved away from Software to the hardware side of the things in terms of providing the persistence / fail-safe storage. If that can be guarenteed all we see in BW is visualisation of data in database; where the database does not really matter; technically speaking.
As for the extractors, what prevents SAP from actually posting data in Memory within the source, it does that similar thing anyways for Logistics.
I wonder what happens to other vendors who are making money because they use extractors as first port of call for SAP adapters in their product designs.
HANA has the potential, how much of that is realised, depends on how things move in hardware and customer adoption space.
Fred Buchanan 03 Jun 2011
Hi Vijay,
15000 would be a huge asset. I've heard that Teradata has ~1000 customers. To be safe let's assume 2000. Based on that they run a business with $9 billion market cap. So you can easily range the BW installed base in that order of magnitude (and put it at least on the same level as the BOBJ and Sybase acquisitions).
How do they monetize on it? Look at BOBJ, look at BPC! Reality shows that it is BOBJ tools being adopted in the previously existing SAP installed base rather than SAP extending its reach into the BOBJ installed base (also: don't rate an SAP customer on the same level as someone who downloaded an evaluation copy of Crystal). Same with BPC: I guess that 2/3 of BPC sales are via the NW version rather than the MSFT one. That basically mimics the same story. BW plays a pivotal part in that game.
And now HANA: I think Vishal & Co. should light a candle every day that BW on HANA will technically work (to then give credibility to the HANA story). If not then the Business Suite on HANA has no chance at all. And then they are left with all the recent demo ware on HANA.
@John: great blog! Thanks for raising this topic and triggering a discussion!
Clinton Jones 02 Jun 2011
I couldn't agree more with your analysis of what it really represents. BW/BI is always going to have a place and a use that is completely different to what HANA will offer on top of ERP.
To assume that analytics in the BW/BI space is the same as the kind of analytics that HANA will excel at, is to over simplify the requirements of business.
I am looking forward to how this will play out over the coming years, I just hope that customers are keen to adopt and that the hardware is able to keep up !
Vitaliy @Sygyzmundovych 02 Jun 2011
John, is it a co-incident that you publish this literally next day after BW 7.3 went into GA? :-) Welcome to the land of walking dead?
May be I do not need to re-post what I have said on few occasions, like here http://bit.ly/fp3Pmt, but the fact is that Vijay's question of "does SAP care if BW lives or not anymore?" bothers me too. Or even more "does SAP care about 9000 customers of BW anymore?" While as we saw during TechEd keynote the number of BW installations grow, we need to see more radical changes than in 7.3. While current release delivered many good integration features, I hardly can call BW 7.3 "the most beautiful release". We really need to hear what is the ultimate vision for enterprise data warehouse product by SAP - performing, agile, but as well delivering state-of-the-art IT capabilities in modeling, administration and monitoring running on different devices. And we need to hear how current BW customers are going to migrate to this ultimate state - I do not believe in perfectly non-disruptive, but at least without rip-and-replace.
Vijay 02 Jun 2011
The fundamental question is - does SAP care if BW lives or not anymore?
BW has always been the ugly one when it comes to customer adoption. What percentage of SAP instal base actually use BW to a good extent? I heard a number like 15000 thrown around. That is probably a single digit percentage of the customer base. Plus, SAP does not make a lot of money selling it any way. The BO suite is the one that brings home the money for them.
With HANA going under BW - will that improve BW's chances?. Sure it will - to some degree. The ones who will be excited are the ones who have BW now. Not sure how many new customers will jump into BW for that reason alone. But the big attraction will be HANA going under business suite - and there by reducing the need for BW to a great degree - and also DS for that reason.
As HW continues its downward trend for cost - OLAP and OLTP should converge even more. And with innovative federation and virtualization of data - BW as we know it will go away. The concept of cubes etc will probably stay as a logical abstraction, but it should become a very low TCO solution. At least I hope that is the case.