Once a run of System Copy GT has been completed then the results can assist in tuning your next run.
Use the Intervals button in Diffuser Mode enables you to see all of the intervals and their runtime. By sorting the runtimes you are able to see which intervals took the longest, if consistently the same table is seen as having the slowest interval runtimes then you may be able to take steps to tune the next run.
The Conversion Logs (see conversion logs section here) can also be reviewed to see the number of entries converted for each table some large tables have no entries to be converted.
Consider the following options:
- Does this table have entries that need to be updated? Were any entries for this table converted in the Conversion Logs? Can it be excluded? Tables can be excluded completely from the run via the SAP standard table BDLSEXZ using SM30 and add an entry with class ‘T Excluded Table’ and object name of the table to exclude (further details are found in OSS note 932032). Or you can exclude from certain logical system names via the selection screen.
- Could the commit size be adjusted, see the Advanced Options section. This is critical as the size of the commit can directly affect the performance of both the standard BDLS transaction and System Copy GT. If this value is set to too low, then far too many COMMIT’s are made to the database – introducing an unnecessary overhead. In addition, too many intervals will be created during the interval generation process, which will also introduce an unnecessary overhead. Conversely, if set too high the commit area can become very large and even cause instability with database interruptions.
- Does the table have a complex key involved? Should you consider the “Dynamic Key” option? Check the Advanced Options section for details as this should only really apply to certain types of tables.
A run with intervals exposed and runtimes sorted is shown as below.
Post your comment on this topic.