The buzzword of the 2012 SAP TechEd was HANA. All sessions were fully booked with long lines in the corridors with IT managers, developers, DBAs and BASIS administrators anticipating the miracles of memory computing performance.
I attended sessions about archiving, near-line and preparedness to uptake HANA. On the other hand, those sessions were almost empty, some having less than 10 people in the room. It makes me wonder: do people realize that HANA may be the future, but to be able to deploy it economically and practically, you need to manage the data volume in your current production environment?
It is very common to hear SAP BW systems reaching 20 Terabytes, 50 Terabytes or even 120 Terabytes in size. But the size of HANA is typically in the Gigabytes range and at a maximum in the low Terabytes of data. This means the BW databases will require a diet before migrating to HANA.
There are only a few solutions available for this diet to happen: archiving, near-line or re-engineering.
The latter means go back to the business and evaluate if all Data Stores (DSO) and InforCubes are really providing benefits for the business and if not, just get rid of them. It may be a year’s worth of work and the end result may not be sufficient to reduce the size of the production database in time for a HANA migration.
Archiving is also an option, but causes concern in the majority of business users, because access to archived data may take a long time, may involve bureaucratic procedures across different groups and IT or simply there is not extra space in the database to reload the data to be accessed. In other words, it is very unusual for customers to archive data in OLAP systems.
The most suitable solution is nearline. Data is moved from production to another location, but the data is always available for queries. Data Stores and InforCubes can be trimmed and all queries can access the trimmed data by literally setting one flag at query level informing SAP BW to pull data from both the live database and the near-line system together.
But what data is a candidate for nearline?
To answer that question, I need to introduce the concept of “data temperature,” which is a subject in itself for another post. For now think of data temperature as: frequently used data and very frequently used data stays in the production database, while infrequently used data is moved to the nearline location. And surprisingly enough, statistics show that not much of the data in InfoProviders are accessed frequently.
Informatica has a certified nearline software product to use with SAP BW systems. The Informatica ILM Nearline software is tightly integrated with the SAP BW system, and all activities to move and read the data is done thorough the SAP user interface, using SAP standard procedures. Informatica ILM Nearline stores near-lined data in an optimized, compressed format, achieving up to 98% compression to dramatically reduce storage requirements, while keeping data accessible at relatively high performance.
Now you can make your BW system leaner and ready for the migration to HANA. And don’t worry: once on HANA, the nearline solution is still compatible and works the same way as if you were running BW on a traditional relational database. It means Informatica ILM Nearline for use with SAP BW is indeed HANA compatible.
In the next few days I will post more details about our solution, with more information about data temperature, architecture and best practices learned from customers using our software for the last eight years.