Introduction
Improve decision-making capabilities by bringing unprecedented performance to SAP NetWeaver Business Warehouse with SAP HANA for current data and SAP Sybase IQ for historical data— achieving the perfect balance of cost and performance. The adapter for Sybase IQ as a near-line solution is delivered with the BW system. Integration of Sybase IQ makes it possible for you to separate data that you access frequently from data that you access rarely, thus making less demands on the resources in the BW system. The near-line data is stored in compressed form. You can thus reduce the costs incurred by data that is accessed less frequently. Please Note that the SAP-NLS Solution can be used with all supported Database Versions supported by SAP NetWeaver BW 7.3x. SAP Near-line-storage for SAP BW is general available with SAP NetWeaver BW 7.3 Support Package 9 and Sybase IQ 15.4
The document https://scn.sap.com/docs/DOC-39627 will guide you through the implementation process.
Some Background Information regarding the new SAP Implementation of the SAP Business Warehouse (BW) Near-Line Storage (NLS) Interface for SAP IQ
The BW NLS interface was introduced already in release BW 7.0, but could only be used thus far in SAP development partner implementations. A number of partners showed interest in this niche and developed a variety of NLS solution concepts.
(click on the image to continue)
One quite remarkable development here was an initiative by the software company PBS from Bensheim in Germany, a longstanding SAP development partner whose main focus is on archiving. In its BW NLS implementation, PBS opted for the analytic database platform Sybase IQ as the storage location for its NLS data, and had very positive experiences with the column-based architecture of this database. Notably, this was prior to the acquisition of Sybase by SAP.
Since BW release 7.3 SP09 is available, SAP is now heavily involved in the database sector, has acquired Sybase and developed its own in-memory database, the legendary SAP HANA DB.
In future, HANA DB and Sybase IQ will work hand in hand as the main cornerstone of the SAP Real-time Data Platform (RTDP)
With this in mind, the NLS strategy in Business Warehouse has also undergone a fundamental change.
To begin with, SAP will take complete control of persistence management for its Business Warehouse and will offer an SAP owned NLS implementation for Sybase IQ. Instead of having to rely on partner solutions, customers will then receive all the components required for running an EDW directly from SAP, including the DWH application (SAP NW BW), the databases (SAP HANA DB and SAP Sybase IQ) and an intelligent NLS management solution.
SAP is pursuing a three-step color-coded persistence management concept.
Active data that has to be accessible on a permanent basis for read and write processes is referred to as ‘hot’ data. As would be expected, this data is stored in the main memory of the HANA DB.
Warm data, which is not accessed quite so often, is best stored in the HANA DB file system. Whenever this data is loaded into the main memory for processing, it is displaced with higher priority once processing has finished.
In future, HANA DB and Sybase IQ will work hand in hand as the main cornerstone of the SAP Real-time Data Platform (RTDP)
With this in mind, the NLS strategy in Business Warehouse has also undergone a fundamental change.
To begin with, SAP will take complete control of persistence management for its Business Warehouse and will offer an SAP owned NLS implementation for Sybase IQ. Instead of having to rely on partner solutions, customers will then receive all the components required for running an EDW directly from SAP, including the DWH application (SAP NW BW), the databases (SAP HANA DB and SAP Sybase IQ) and an intelligent NLS management solution.
SAP is pursuing a three-step color-coded persistence management concept.
Active data that has to be accessible on a permanent basis for read and write processes is referred to as ‘hot’ data. As would be expected, this data is stored in the main memory of the HANA DB.
Warm data, which is not accessed quite so often, is best stored in the HANA DB file system. Whenever this data is loaded into the main memory for processing, it is displaced with higher priority once processing has finished.
To achieve this, the BW development team has developed a non-active data concept in order to optimize the displacement strategy for BW tables in the HANA DB. Here, tables with warm data are flagged and are prioritized for displacement from the main memory whenever they are used. These tables are normally partitioned in the default setting, and only the partitions of the table that are affected by the transaction are loaded to the main memory.
This makes it possible to make more efficient use of the priceless commodity that is main memory space.
This makes it possible to make more efficient use of the priceless commodity that is main memory space.
Differentiations: Multi Temperature Data Approach
While the non-active data concept manages the passage of warm data between the main memory and the disk, NLS management is concerned with the next data layer. This is ‘cold’ data that is rarely used and for which no more updates are anticipated in BW. This data is removed in time slices to the Sybase IQ DB. All affected time intervals of an InfoProvider are locked for load processes in BW.
Apart from this restriction, the removed NLS data remains available at any time for all other processes in BW, for read accesses quite naturally but also for ETL processes or individual look-ups.
Strategic queries, which analyze data over longer periods of time and thus inevitably also include cold data, are split and procure data both from HANA DB and from Sybase IQ.
The removal process itself is highly complex and must comply with transactional requirements. All this can be easily managed in both directions using BW Data Archiving Processes (DAP) which can be combined in Process chains. Finally, the consistency of the entire data pool must be ensured at all times, even if this is split into two disjoint parts in the case of NLS.
The interface to the Sybase IQ database required for this was implemented by SAP developers in Walldorf, on the basis of the existing NLS interface and its SDK.
Near-Line Storage (NLS) Architecture - LibDBSL based
On the basis of the ODBC connection to IQ, and separate porting in NetWeaver, this makes it possible to create a connection in ABAP from BW to IQ. With this connection, Sybase IQ Loader technology can be used for extremely high-speed writing of the BW data to be extracted to the NLS tables on the IQ.
Initial tests show that this can be done at a rate of about 200 million sets of data per hour. With these figures, SAP is way ahead of its competitors.
Initial tests show that this can be done at a rate of about 200 million sets of data per hour. With these figures, SAP is way ahead of its competitors.
Find more details in the recorded Live Expert Session (LES) and SAP NetWeaver BW Near Line Storage (NLS) Implementation for Sybase IQ
Introducing SAP HANA dynamic tiering
HANA's dynamic tiering capability is shipped with HANA SP09. In combination with BW 7.4 SP8, it substitutes the pilot setup described above. There will still be extended tables but the separate IQ server is replaced by an ES server that is integrated in the HANA installer, update, management, back-up and recovery mechanisms. There will be a designated node on the HANA hardware running that ES server. This setup will be released for general availability (GA). There will be no migration options from the pilot to the GA version. Details can be found in a PDF presentation attached to OSS note 1983178 or the online documentation.
No comments:
Post a Comment