Log Manager window

Use the Log Manager window to check Read and Write activity both to and from the log.

To display this window, select Log Manager from the DB2® Statistics Detail panel.

Figure 1. Log Manager window

 DGOMSWLG                   Log Manager

                                            INTERVAL  2:48.421
                                                       More:   - +
 Reads satisfied
   From output buffer  . . . . . . . . . . . . . . . . . . :      0
   From active log . . . . . . . . . . . . . . . . . . . . :   1475
   From archive log  . . . . . . . . . . . . . . . . . . . :      0

 Reads delayed
   Tape volume contention  . . . . . . . . . . . . . . . . :      0
   Unavailable resource  . . . . . . . . . . . . . . . . . :      0

 Look ahead mounts attempted . . . . . . . . . . . . . . . :     23
 Look ahead mounts successful  . . . . . . . . . . . . . . :      3

 Write-no-wait . . . . . . . . . . . . . . . . . . . . . . :    151
 Write output log buffers  . . . . . . . . . . . . . . . . :      3

 Total BSDS access requests  . . . . . . . . . . . . . . . :     26
 Unavailable output log buffers  . . . . . . . . . . . . . :      0

 Active log
   Control intervals created . . . . . . . . . . . . . . . :      3

 Archive log
   Read allocations  . . . . . . . . . . . . . . . . . . . :      0
   Write allocations . . . . . . . . . . . . . . . . . . . :      0
   Control intervals offloaded . . . . . . . . . . . . . . :      0

  Output log buffer writes per
   active log control interval created . . . . . . . . . . :    1.0

  Command ===> ____________________________________________________
  F1=Help      F2=Split     F3=Exit      F5=Auto      F6=History
  F7=Up        F8=Down      F9=Swap     F10=Delta    F11=Interval
 F12=Cancel   F16=Look     F17=Collect

If a high percentage of reads comes from the archive log, it might be necessary to increase the size of the active log. For example, a large update job with few commits could fill the active log forcing an archive. If the job fails, recovery is required to retrieve records from the archive log. Archive activity can be expensive in terms of response time, especially if the archive log is placed on slow devices such as tape or cartridge.

The following list describes some of the important fields that are shown in this panel:

Reads delayed - Tape volume contention: Number of read accesses delayed because of tape volume contention (that is, a tape volume was already in use by another thread).
Reads delayed - Unavailable resource: Number of read accesses delayed because of an unavailable resource. This can be because of an insufficient number of tape units allocated, or because the archive log read service task is not available.
Write output log buffers: The number of Write requests issued irrespective of single or dual logging. This field is updated once per buffer Write. The update value is either one or two I/Os, depending on which logging option is chosen (single or dual). This should have a value consistent with the known workload update rate.
Unavailable output log buffers: This field shows how many times a Write request to the active log had to wait because no buffer was available. The value should ideally be zero as these waits should not occur. If these waits do occur, the output buffer might be too small, or the size of the write threshold might be too close to the size of the output buffer.
Active log - Control intervals created: Number of active log output control intervals created. Log records are placed sequentially in output log buffers, which are formatted as VSAM control intervals. The control intervals are written to a set of predefined active log data sets, which are used sequentially and recycled.
A useful ratio is: Write output log buffers divided by Active log - Control intervals created.

Logging Considerations

Minimize device contention on the log data sets by placing data sets correctly. If you use dual logging, do not place both logs on the same volume.
Avoid waits that occur because no log buffer is available.
Define enough active log data sets to prevent Db2 from waiting while a log is archived.
Make the active logs large enough that backouts do not have to use the archive log.
Consider the 3990 DASD FAST WRITE controller for the log. Performance measurements have shown that sequential access mode with DASD FAST WRITE provided substantially better performance than native DASD when the amount of log data written per commit was 24 KB or less. DASD FAST WRITE performance was comparable to that of native DASD when 48 KB of log data was written to DASD for each commit. When more than 48 KB was written, native DASD performed better than DASD FAST WRITE. Therefore there might be a need to determine in which environments log performance is critical to assess the value of DASD FAST WRITE.