SAP promises that in just a few days, "Analytics will never be the same again". With that, the BI community is waiting in anticipation of what will be revealed.
There is no doubt in my mind that the latest addition to the SAP HANA Cloud Platform will be a state-of-the-art BI solution. There is also no doubt in my mind that very soon after launching SAP Cloud for Analytics, people will ask the question again: Will HANA replace BW? The answer to this question is slowly changing. The latest update was written in July this year, suggesting a ‘no, but…’
The answer this time will still be a ‘no but…’ The SAP HANA Platform is still a few iterations away from being a full-fledged Enterprise Data Warehouse (EDW) Platform. The gap is closing though, so let’s have a look at what the SAP HANA platform has to offer in the context of Enterprise Data Warehousing, and what the shortcomings are.
For this post I have imagined an EDW solution based on existing functionality on the SAP HANA platform and on speculation of the functionality which will soon be delivered by the ‘Cloud for Analytics’ offering. Whether you can indeed seamlessly integrate the ‘Cloud for Analytics’ functionality in an existing HANA implementation or if you can use all HANA platform functionality within the ‘Cloud for Analytics’ solution, I don’t know. I expect there will be some caveats to start with but I also expect that soon all analytic functions offered in ‘Cloud for Analytics’ will also be available on the HANA platform.
Source system connectivity
There is a rich variety of functions available on the SAP HANA platform to integrate data from different systems, either virtually or physically. SAP BW has in fact recently made a leap forward in this area by leveraging from HANA platform functionality – connectivity has not always been SAP BW strongest point.
One caveat is that the tools for integrating data have always been in the hand of the BI team, not the business user. Business users would clearly benefit if they could add data from other systems into their reporting area themselves. There is a framework for this in BW (‘Work spaces’) but it is not terribly user friendly.
SAP now promises to allow end-users to blend off system data with enterprise feeds which, if it is truly user friendly, will have a warm reception in the BI community.
As it currently stands, there is no doubt in my mind that the SAP HANA is second to none in delivering the system connectivity function in an EDW scenario. Adding end-user functionality for data blending will only reinforce this position.
Defining and applying business rules to build an integrated, non-ambiguous dataset for reporting
Re-usability of rules and definitions help to build a consistent data set. There is an element of re-usability of components of virtual models on the SAP HANA platform (using HANA analytical views) and rules defined in Smart Data Integration can be used in different transformations within SDI. There is no metadata repository or an administration of business rules (owned by the business) so here the solution is somewhat incomplete.
Additional tools, like SAP Information Steward, could be utilised to provide this function. Again, BW leverages, to some degree, from HANA functionality and has a more comprehensive framework for master data administration. This puts BW in a slightly better position, but ultimately it would be fair to recognise that both within the HANA Platform and within BW there is room for improvement. I doubt that ‘Cloud for Analytics’ will change this.
Data Stores and Data Staging
BW has a whole rafter of object types which can be used for standard use cases within an EDW. There are delta-enabled staging objects, master data objects with hierarchies and time dependencies, the concept of non-cumulative key figures, write optimized objects and so on. None of this is available as standard objects on the SAP HANA platform. Most of these objects are still relevant (I have deliberately left out ‘Cubes’) so I do believe the HANA platform has some catch-up to do. Already some functionality for defining data stores and staging objects is moving from the BW ABAP GUI to the Eclipse environment. At this point in time, you still need to run BW to be able to use this functionality in Eclipse, I expect we are not miles away from having this functionality available on the HANA platform without the need to run BW. BW is a best of breed product for data staging in an EDW scenario so it shouldn’t be too difficult for SAP to replicate this to the HANA platform. Again, I don’t expect ‘Cloud for Analytics’ to make a difference in this area.
Process automation and monitoring
SAP HANA started as a database, and over time different functions have been added to the platform. With the introduction of Smart Data Integration (SDI) the HANA platform suddenly had a state-of-the-art ETL tool on board. SDI is great at connecting different sources and transforming data using complex transformation logic. What it does not do very well, is providing a monitoring function for scheduling and monitoring batch processes. This is a shame, really, because it is one of the things that BW does very well. It is indeed hard to find fault with the BW tools for scheduling and monitoring. Without an improvement on this front, it will be difficult to imagine anyone would run an EDW on the HANA platform, without a third-party tool to provide the scheduling and monitoring function.
The data modelling capability of BW has always been a double edged sword. The use of InfoObjects as building blocks for master data and DSO’s and Cubes has been great for consistency and re-usability. It does add some extra development time to each and every process though and this is causing some frustration when doing a proof of concept or a prototype. With BW on HANA, it is now possible to skip the step of defining InfoObjects and create a ‘field based’ model instead. This is incredibly helpful and certainly speeds up development. On the HANA platform, without BW, there is no real data modelling function other than modelling of analytical models. Clearly there is a gap here so it is no surprise that ‘Cloud for Analytics’ now promises ‘Agile Data Modelling’ and even the option to ‘integrate models and data definitions from BW’. So is it the best of both worlds? I certainly hope so. It would give us different tools for different use cases and that would make it a very attractive solution.
The obvious tools to talk about are the core BI tools for visualisation, data exploration and pixel perfect reporting. But it is important to understand that an EDW should cater for other use cases as well, such as predictive analytics, planning applications and text analysis. BW is more and more seen as the back-end platform, fulfilling a data provisioning role for other tools. The tools which are integral part of BW are based on old technology and not further developed (BEx reporting, Analytical Process Design, Integrated Planning). There is no shortage of interesting tools for core analytics, predictive analytics or planning which can work on top of BW (The BOBJ toolset, BPC). The HANA platform has more to offer though: It comes with predictive engines, text engines and now also tools for visualisation, analysis and reporting. How Cloud for Analytics holds up against best of breed tools like Design Studio, Analysis for Office, Lumira and Webi remains to be seen.
BW has an authorisation model which is based on business roles, allowing easy control of who has access to what level of detail in specific data sets. The HANA platform does not have this yet. The current HANA framework for security is a mainly a standard database authorisation model, with a very light layer of Analytics security on top of it. What is currently missing is an authorisation model which is tailored around deploying the HANA platform as an EDW application.
A couple of years ago, the question of whether SAP HANA was replacing BW was answered with a firm ‘no’. HANA is, and was, a database and a high performance analytical appliance. Back then it lacked several of the important features BW – or any mature EDW solution – has to offer. Things have moved quickly and new functionality is added to the HANA platform every few months. HANA was not set up to be an EDW platform, it is much more than just that. But it is certainly starting to look like you can utilise the HANA platform as an EDW. The main shortcomings I can see, as discussed above, are around process automation and monitoring, data modelling and staging and security. These are all functions which are provided in BW, so BW on HANA makes an excellent EDW platform, but HANA on its own is not there yet.