Data Tiering Optimization with SAP BW/4HANA
Introduction
The data storage optimization is an emerging topic, which is becoming more and more important within organizations. A detailed analysis is necessary to avoid the losses that data storage can generate.
How to maximize performance and optimize data storage while minimizing total project costs?
It is highly recommended to implement a data aging strategy in BW systems, due to the large volume of data and its continuous increase.
The adoption of this strategy improves the performance of the BW System, classifying the data by access type, which are stored in different areas of memory.
The implementation of a multi-temperature strategy classifies data by frequency of access into hot, warm, and cold. These frequencies are detailed below:
Hot Data
- Frequently used data.
- Reporting data.
- Data is stored in SAP HANA.
Warm Data
- Not frequently used data.
- Harmonization and consolidation data.
- Data is stored in SAP HANA.
Cold data
- Historical Data.
- Used for data archiving.
- Data is stored outside SAP HANA.
Diagram representing the Data Lifecycle. © 2021 SAP SE or an SAP affiliate company.
Data Tiering Optimization: What's this?
Data Tiering Optimization or DTO supports the classification of data in the advanced DSOs like mentioned above. It is an option to optimize the memory footprint of data in SAP BW/4HANA.
DTO provides a central UI to configure data storage options according to the temperature to be set. The temperature of the data is allocated by partition in the aDSOs.
Explaining the data tiers. © 2021 SAP SE or an SAP affiliate company.
Before implementing DTO
Before implementing the DTO it is suggested to execute some takes like housekeeping tasks to keep the system more performative:
Whenever you have a big inbound table, you should check whether activation compression is possible, especially with ADSO type DataMart. When you have big actives table it is recommended to use of BW / 4HANA data management (HOT, WARM and COLD Storage in DTO). And for the big changelog table keep up with the housekeeping.
Moreover, some important aspects should be taken into consideration, like the number of records on GoLive, the expected growth in the future, and possible partitioning characteristics. A later conversion of the data model to partitions is possible, but time-consuming and involves the risk that remodeling processes will abort.
Best Practices – Partitioning of ADSOs
These are some of the main recommendations from SAP:
Even if the aDSO holds more data, the number of data records in one without partitions should not exceed the maximum of 1 billion.
It is important to emphasize that partitioning has no negative influence on data management (loading of the data), rather the system benefits from faster delta merge operations (activation of the data).
SAP also recommends using time characteristics to be used in the partitioning (0CAL * / 0FISC *) because these provide optimized access to the partitions during query runtime. Only the partitions that contain relevant data will be read (DB pruning concept).
The partitioning element should be included in the query design (with a variable or a fixed filter in the global).
According to SAP experts, for the reporting performance, the use of partitioning has a small positive influence (10-20%), if DB pruning can be used.
Conclusion
By using DTO with SAP’s recommended best practices, we have a high performance BW system with efficient memory management, reducing main memory usage which directly impacts the project cost.
Do not hesitate to contact us at info_twca@temwork.net for more information!
