Log Apply Performance Considerations

Review the following considerations to optimize your Log Apply jobs for the best performance:
caladb2
Review the following considerations to optimize your Log Apply jobs for the best performance:
Parallel Processing
When you generate a load file, you can write the load output to a single data set or to multiple data sets. The use of single or multiple data sets does not affect performance significantly. However, you can achieve some degree of parallelism and performance enhancement by taking the following actions:
  • Specify multiple output data sets.
  • Separate the single Log Apply control statement data set into multiple control statement data sets, each with a separate Log Apply control statement.
  • Submit the jobs simultaneously.
Multiple Row Inserts
When performing the insert-only mode, set the pla_format parameter of LOG APPLY to
M
. This setting enables the insert-only mode with multiple row insert (MRI). MRI decreases the number of accesses to DB2 and saves the CPU time and elapsed time compared to insert-only mode (pla_format = I). The pla_mrinum parameter specifies the number of rows inserted by one MRI statement.
Additional storage requirements need to be considered. Use the following formula to calculate storage requirements for MRI processing:
(r1*pla_mrinum+2*s1) + (r2*pla_mrinum+2*s2) + … + (rn*pla_mrinum+2*sn)
  • n
    Specifies the number of tables in control file.
  • r1
    Specifies the length of rows in the table.
  • s1
    Specifies the number of columns without NOT NULL option in the table
Allocate the resulting number of bytes in addition to the standard amount that Log Apply uses in the inser-only mode using the same input data.