The most important post-installation activity after a system copy is running BDLS, which converts the logical system name from the source system name to the target system name. This is also the longest running process which increases the overall System copy process duration/time.
Running BDLS on Pure Storage is again fairly simple and will not require you to do any kind of fine tuning as long you have enough background jobs to achieve parallelism of running BDLS on Big tables. Pure Storage FlashArray//m comes with high bandwidth (SAPS) which should be tapped to the maximum for accelerating the BDLS execution.
In order to minimize the BDLS execution time, use the following approach. This approach would make sure that whole BDLS process can be parallelized easily with the help of ABAP programming.
Here are the steps to parallelize BDLS process to the maximum:
- Increase the number of Background jobs in transaction RZ10 by changing profile parameters for background jobs. This is based on the number of cores your application server supports.
- In the BDLS transaction create variants for each of the set of big tables. You should split the number of tables equally among the number of background jobs.
- In the transaction, SE38 create a new Z program. In this program create a standard table and load all these variants into this standard table.
- Using function JOB_SUBMIT loop through the above standard table submit the BDLS jobs. The name of the BDLS program is “RBDLSMAP” which should be used in the JOB_SUBMIT report parameter.
This program run should create many background jobs for each of the variants. Using this method I was able to generate 40000 SAPS by using 10 background jobs. This approach should accelerate your BDLS process by 3-4 times on Pure Storage FlashArray//m.
This program can be downloaded from Github