SAP HANA – In-Memory-Technology
// Business Intelligence, SAP-Processes & Systems
As easy as Apple, as fast as Google?
In-Memory Technology currently is one of the most interesting technical topics in information technology that will surely bring about several changes in the near future. At present, its main area of application is analytical systems that achieve an enormous performance gain from using In-Memory Technology. Additional areas of application are presently being tested and piloted at large manufacturers. Here, SAP is taking a very systematic, innovative path with its In-Memory solution HANA.
Companies at which SAP software is playing a central role in the application landscape will have to deal with this topic sooner or later. Quite often there is currently considerable uncertainty about how to deal with SAP HANA. Is it already ready for the market? Will it play a role for your own company? When is the correct point in time to develop your own activities? What does your own strategy look like? Some information for a better understanding and some suggestions may help to get closer to the HANA topic.
At first, some pointers regarding the foundations of the HANA technology. From a system technology point of view, SAP HANA is based on a hybrid In-Memory database. Said database combines row, column and object-based database technology that has been optimised for the parallel processing functionality of modern multi-core / CPU architectures. For this, special hardware is needed which is provided by SAP partners. All of this is then running on a SUSE Linux Enterprise Server (SLES). But SAP HANA does not only represent the system environment just described, but rather has to be viewed as a package that also provides corresponding software in addition to the system environment. Depending on the scope, said software is delivered in different editions. The most important component, in any case, is the database which in a way represents the interface to the system environment. Additional HANA components exist around it; some of the important ones are:
The most efficient way to load data into SAP HANA is via replication.
As the illustration shows, there are different types of replication (trigger, ETL and log-based replication) which, however, will not be covered in more detail, here.
SAP HANA constitutes a complex system and application environment which will bring major changes in the application architecture with it. One essential change will be that SAP ERP and SAP BW will run on a central HANA database. This will not happen from one day to the next, but will rather be an evolutionary process. For a company, to be able to adjust to SAP HANA, it is necessary to understand this process. Let us therefore take a look at the HANA Roadmap of SAP. The development process takes place in roughly three phases.
In Phase I („SAP HANA 1.0"), the HANA database was developed as a separate database. Data could only be transferred to the database via replication; database application could be created based on it. The area of application was limited; this phase has been completed.
In Phase II („SAP HANA 1.0 SPS3"), HANA serves as primary database for SAP BW. At present, this is the current state; in this phase, a „real" integration of SAP BW and HANA is taking place for one, existing functionality is shifted from SAP BW to HANA (e.g. activation of DSO objects), for another, new functionalities are created that are only available on a HANA platform (e.g. Workspaces). Further below, we will address the essential changes that result from this for SAP BW in more detail.
In Phase III („SAP HANA Vision"), HANA serves also as data basis for the ERP, as a result of which, the ERP and BW will be running on one and the same platform. This phase will be achieved in the mid term, most likely within a period of two to three years. It is anticipated that this phase will have the strongest impact, in particular with respect to the topic of Business Intelligence. Which impacts those will be specifically, depends on whose imagination is being tapped into, a precise forecast is rather difficult at this point in time.
Let us first take a look at which impacts the current state of development of HANA has on SAP BW. With the integration of SAP BW and HANA, BW is the first application that effectively utilises the strengths of SAP HANA.
- The essential effect is, of course, the enormous increase in performance, for one when executing queries, but also when in case of administrative tasks such as the activation of DSO objects.
- Due to the high compression rate, the data volume is significantly reduced
- The BWA (BW Accelerator) will become obsolete
- Significant performance increase of the ETL processes (DSO and Infocube)
- Streamlining of the BW modelling due to utilisation of
- HANA-optimised DSOs
- HANA-optimised Infocubes
- Integration of ad hoc models that offer the specialist departments a simple, quick opportunity to integrate information that does not (yet) exist in BW, e.g. aggregation criteria for planning or simulation. This can be done via Workspaces which presuppose a SAP BW / HANA.
- Through HANA-optimised planning, the planning process can be optimised and designed with higher performance.
It is important to understand how BW modelling will change due to HANA. Here is where the biggest different is, e.g. in contrast to the utilisation of a BW Accelerator of also to the utilisation of HANA as a side-by-side solution. Even though modelling is in principle simplified, an exact knowledge of HANA-optimised modelling is necessary, nevertheless. Approaches for a HANA-optimised architecture are on SAP‘s part described via the LSA++ architecture, which will not be discussed further here.
With respect to the other impacts that a future integration of ERP and BW on a single HANA platform will bring along, only general statements can be made since - naturally - there is a lot of leeway for speculation. Statements that BI systems will become superfluous this way and that through the performance potentials of In-Memory Technology, operative data can be reported on directly may be missing the mark, just like assumptions that the separation between ERP and BW system will remain in its current form. Fundamentally, a separation between analytical and operative models will remain. But this separation, as a tendency, will likely be achieved via virtualisation („Analytical Views"). ETL processes in the conventional sense will then likely be a thing of the past. This development will have a strong impact on the type of modelling as well as on the options resulting in the area of real-time analyses. From today‘s point of view, it is important to recognise the tendencies of these changes and to monitor their further development.
At present, the integration of HANA and SAP BW represents the most important application case. In addition, there are other ones that will not be listed here in detail, however. One application case, but one that could play a role for a lot of current SAP customers, would be the HANA / CO-PA solution. The classic CO-PA is based on relatively simple data structures which then, however, are often faced with an enormous data volume. Quite often this leads to a situation where the processing times in CO-PA become unbearably long, or that planning processes within CO-PA very much lack performance. Here, SAP is offering an out-of-the-box solution that roughly works as follows:
The COPA data of the ERP is replicated into the same structures of a HANA database. An additional layer in the ERP („CO-PA read function module") decides which data source is best for a specific functionality; in general, this will be HANA. This out-of-the-box solution can be a sensible or even necessary solution for companies which, however, will lose in importance over the course of the development.
HANA is strategically the platform for ERP and BW. Consequently, all SAP-based companies will sooner or later have to deal with this technology. Even though the statement „As easy as Apple – as fast as Google" may be applicable at its core, the path to get there is not always an easy one for companies, especially since HANA is more than just an In-Memory database. You have to ask yourself
- how mature the technology is. That it is available does not necessarily mean it is usable.
- how large the pressure to act is at the company.
- whether an SAP (BW) landscape already exists or is going to be set up.