From 650 to 25 GB in One Go

Company profile RI Solution
Company profile RI Solution

The volume of business-critical data is growing rapidly, because enterprises need more and more detailed information from all areas of the company. This profusion of data needs to be brought under control, that is, stored and made available to users. It is something that the IT service provider RI Solution GmbH used to struggle with. The explosive growth in the volume of data in the SAP systems hosted for customers not only required a ever-increasing amount of storage space, it also adversely affected system performance in day-to-day work such as reporting or the building of statistical data and aggregates, which are redundantly stored subsets of an InfoCube, usually with a low granularity.
RI Solution runs SAP R/3 and SAP NetWeaver – Business Intelligence (SAP NetWeaver BI, the former SAP Business Information Warehouse, SAP BW). Over time, however, data storage for SAP NetWeaver BI became difficult, as Bardo Gutberlet, head of SAP system development and support at RI Solution, explains: “While there were archiving functionalities for SAP R/3 from the very beginning, an adequate solution did not exist for SAP NetWeaver BI. Over the years, the number of SAP NetWeaver BI users among our customers grew to around 1,300. The size of the database totaled three terabytes.” One terabyte of this was made up of data from the PSA (Persistent Staging Area), which customers must retain permanently online as they need original data for historical analyses.

Long runtimes for backup and recovery

“On top of the huge costs for storage, we also had problems with regard to administration. And in addition, the long runtimes for queries or backups were a headache,” says Gutberlet. Work such as the change run, for example, which is the adjustment of aggregates in line with master data changes, was only carried out on the weekend, in order to avoid disruption to day-to-day business. For reporting and analysis, this meant fewer and fewer current assignments to master data attributes over the course of a week, and the resulting risk of misinterpretations.
The runtimes for backup and recovery also continued to increase, with the backup of the SAP NetWeaver BI systems taking a good 12 hours. It was particularly difficult to restore the large datasets in the PSA area used in SAP NetWeaver BI from the SAP R/3 source systems, because of the long loading times of up to several days. This activity can only be carried out at times when no postings are made and in some cases, the data could not be restored at all because it had been deleted in the source system. But this information was still needed to build new, application-specific InfoCubes that met additional reporting requirements.

Stemming the growth of data

The situation was untenable. The RI Solution specialist department responsible for operating SAP NetWeaver BI needed an efficient solution for archiving and restoring InfoCubes, PSA and ODS Objects (Operational Data Store). At the same time, more flexible handling of the archived data was needed, as PSA objects for example still need to be available as information sources. The company did not want access to the archived data to result in poorer performance for users. “We had to stem the growth of data in SAP NetWeaver BI to guarantee the service level agreements for our customers,” explains Gutberlet.
In the summer of 2005, therefore, the company started to look for a suitable archiving or nearline storage solution. The latter is the level between an offline archive and a relational online store. In it, rarely used information is deleted in SAP NetWeaver BI but is still available “nearline” for access from the BI solution. Following a market analysis, RI Solution looked more closely at three different products: a classical archiving software, a transactional database enhancement, and the nearline storage solution SAND/DNA Access from SAP partner SAND Technology. Following a test installation, the enterprise chose SAND/DNA Access.
The implementation started at the end of April 2006 and was completed in October 2006. Responsibility for the project lay with a small team consisting of one consultant from SAND available on call and two employees from RI Solution, who could only work part-time on the project because of their day-to-day work. The implementation of the software took just a few hours. In this process, the SAND database was installed on the same hardware as the SAP system. Integration in SAP NetWeaver BI also went smoothly.

Compression rate of 96 percent

The project has paid off. SAND/DNA Access provides RI Solution with the ability to highly compress selected data without restricting its availability for analyses or as the basis for new ODS objects or InfoCubes. “The original volume of data that we have moved to the nearline store so far is about 650 GB. In contrast, the size of the nearline store is around just 25 GB. We have therefore achieved a compression rate of 96 percent,” states Gutberlet.
Even in the tests, the InfoCubes could be maintained more effectively. The lower data volume resulted in shorter total runtimes for standard jobs such as the building of indexes for statistical data or aggregate maintenance with roll-up, that is, aggregate maintenance after the data has been loaded, and the change run. The company has also benefited from the archiving of PSA data: new InfoCubes can be built far more efficiently and quickly, access to current PSA data tables takes less time than before, and reporting on PSA data is possible for the first time. Errors can also be corrected more quickly and the administration costs have fallen to a minimum. The backup of the SAP NetWeaver BI systems now takes just seven rather than 12 days.

Lean processes for data management

“With SAND/DNA Access, response times are not a problem for us any more. We have now started to move InfoCubes, for which response times are even more important than for access to PSA data, because users access them directly. But we are sure we will be able to meet our customers’ requirements here, too,” says Gutberlet.
His conclusions are extremely positive: “SAND/DNA Access is ideal for our requirements. The solution significantly reduces the data volume in the active SAP NetWeaver BI. Our data management processes remain lean thanks to the high degree of integration between SAND/DNA and SAP NetWeaver BI. At the same time, though, the relevant data is not deleted or even transferred to an offline archive, and is still available at short notice. We can access it at any time, and thus meet our specialist department’s demand for information that can be flexibly prepared.”

Dr. Michael Hahne
Dr. Michael Hahne