Case Study: Near-Line Storage System
Makes Aged BW Data More Easily
Accessible
By Mel Mele, Principal consultant, FileTek, Inc.
It is possible to archive and retrieve infrequently accessed data (what SAP calls "sleeping" or "aged" data) without the intrinsic limitations of traditional archiving systems, in particular bulk loading before querying that might require manual intervention. I know this is feasible because I participated on a team that launched a successful pilot system to accomplish just that. What follows is an account of this pilot - the first near-line data storage (NLS) system integration with SAP BW. (If you are unfamiliar with the NLS concept, see the sidebar, "NLS: Taming the Sea of Information.")
I'll show you the various issues considered and logical steps taken to deploy a successful NLS system with BW. If your organization is facing escalating data growth, especially with aged data, our experience with this pilot will give you insight to devise your own strategy for effectively managing large amounts of data. For organizations that have not reached a data deluge, this article will introduce an alternative method for future data management strategies.
Entergy Pilot Overview
Entergy Corporation, a major energy utility based in New Orleans, Louisiana, believes
that over the next three years, it will be managing about 10 terabytes of data.
The company's data architecture team wanted to take a proactive approach to deal
with this anticipated massive increase in BW data. Entergy's plan was to keep the
most recent six months of ODS data in an Oracle database and move the rest to its
NLS system.
A team of engineers from Sun, BIS, FileTek, EDS, SAP, and StorageTek collaborated with Entergy's IT professionals to prove that they could successfully manage massive amounts of aged data without using a conventional data archiving system. The group's main objective was to move aged data from the BW primary database to a secondary relational database management system (RDBMS) that supports economical storage media (in this case, tape) and provides direct access to detailed data using standard SQL queries. The group intended to accomplish this goal without reloading files into a new data target, which typically occurs in most archiving systems
The pilot project required three weeks to set up, test, and document by a team of three consultants. An enterprise wishing to deploy an NLS system should estimate around 90 days to plan, design, define a data-aging strategy, and then code and implement the solution for the first business area. Each additional business area should take approximately 60 days each to implement. Table 1 shows the time required for each key step
Table 1 Time required to develop, configure, integrate, and
implement an NLS
| Activity | Staff Days |
|
Develop BW data-aging strategy and policies |
25 |
|
Configure StorHouse server and integrate with BW |
5 |
|
Implement by business area
|
60 |
Design BW data migration processes using
the SAP APIIn the context of the Entergy pilot, the NLS system was a database extension to the BW RDBMS. It combined FileTek's StorHouse® storage and RDBMS software with alternative storage (tape) to provide economical and quick access to large amounts of data. The pilot system had the ability to access detailed data on any type of media through standard SQL queries without first restoring that data to BW. BW always kept track of the data offloaded to the pilot system because mySAP BI objects in the database extension system were registered in the primary database metadata repository to provide a single view of all data.