With these proactive steps, your Remote Conversion project will not only perform better but also ensure a smoother migration experience.
Adjusting Package Sizes
Migrating in smaller chunks can often lead to excessive monitor entries and delays. Instead:
- Set package sizes from 20MB to 120MB. This adjustment reduces the number of entries in the CNVLTRL_TASK table, improving system performance and monitor availability.
Leveraging Parallel Reads
Parallel reads help eliminate bottlenecks during data migration. To achieve this:
- Try to use the first field as the parallel read field: If this is not possible, create a secondary index on the selected field. (for non HANA DBs).
- Check the CNVLTRL_PORTION table to make sure that the portions are equally distributed: If not, use another field.
Use multiple clusters
Aim for about 20 clusters by adjusting the cluster size.
- Use Simulation mode: Utilize it to identify the optimal size for clusters before initiating the creation.
- Node distribution: Ensure that clusters are evenly distributed across nodes on both source and target systems.
Use Filters Strategically
Filters can reduce the data volume during migration, significantly improving performance.
- Use fields like fiscal year or company code as filters.
- For more control, add filter entries through transaction CNVLTRL_FILTER.
- Apply these filters before running "Create Sequence to Generate and Execute Reports".
- Ensure that your filtering strategy aligns with the table's structure and business logic.
Pre-Partitioning for Large Tables
For E- and F-tables with over 2 billion records:
- Pre-partition these tables on the target system based on known entry counts.
- This minimizes the risk of performance degradation during data writes.
Schedule Delta Merge Jobs
Large tables can significantly slow down migrations if not handled properly. Schedule delta merge jobs every two hours on the target system during the insert phase.
Key areas to focus on include:
- Cluster tables.
- Big tables.
- CNVLTR_TASK table.
- Log mode changes: Switch the log mode to "overwrite" on the target system to reduce log storage issues.