Analyzing DEDB areas at the UOW level

By analyzing the RAP CI usage, the free space distribution, and database record length distribution at the UOW level, you can obtain information about the status of the DEDB area in more detail.

Before you begin

Ensure that you can access either of the following reports for the subject DEDB area:
  • UOW report
  • UOW Exception report

About this task

The following two lines show the column headers in the UOW report and the UOW Exception report:
Figure 1. Column headers in the UOW report
          ---NUMBER CI'S USED---  #RAPS USE  ---%RAP FS---   % DOVF    #DB     ------------ RECORD LENGTH STATS ---------------
UOW NO.    #RAPS   #DOVF   #IOVF  OVFL CI'S  AVG  MAX  MIN  PART FS  RECORDS          AVERAGE          MAXIMUM         MINIMUM
            (1)     (2)     (3)      (4)     (5)  (6)  (7)    (8)      (9)             (10)              (11)            (12)

The first column, UOW NO., contains the UOW number, and other 12 columns contain information about the UOW.

The UOW report provides the following information for each UOW that is defined for the area:
  • The number of RAP CIs in the UOW that contain data (column 1)
  • The number of DOVF CIs that contain data in the UOW (column 2)
  • The number of IOVF CIs that are allocated to the UOW (column 3)
  • The number of RAP CIs of the UOW that contain a pointer to segment data in either the DOVF or the IOVF section (column 4)
  • Statistics about the average, maximum, and minimum percentages of free space in RAP CIs that are used in the UOW (columns 5, 6, and 7)
  • The percentage of free space in the DOVF section of the UOW (column 8)
  • The number of database records in the UOW (column 9)
  • Statistics about the average, maximum, and minimum lengths of the database records in the UOW (columns 10, 11, and 12)
Note: The information in the last two list items is reported only when the utility that generates these reports runs in specific run mode.

You can use the UOW report to monitor and manage space utilization at the UOW level. The report helps you determine if the database records are randomized so that RAP CIs and the space inside RAP CIs are used effectively in each UOW.

The contents of the UOW Exception report for a UOW is the same as that of the UOW report. However, the UOW Exception report is printed only when certain utility options are specified. The report contains information about those UOWs that satisfy specific conditions.

You can use the UOW Exception report to simplify the following tasks:
  • To identify which UOWs require reorganization
  • To list only the UOWs that match the predefined space utilization criteria

For the utilities that can generate these reports and necessary utility options, see the topic "Generating the analysis reports" in the IMS Fast Path Solution Pack: IMS High Performance Fast Path Utilities User's Guide.

Procedure

Complete the following steps for each UOW that you want to obtain detailed information about:

  1. Evaluate the efficiency of the randomizing routine that is being used.

    Inefficient randomizing can be identified by the data in columns 1 through 3, which are under the heading NUMBER CI'S USED.

    The #RAPS column (column 1) indicates the number of root anchor points in the UOW, to which at least one database record is assigned. The #DOVF column (column 2) and the #IOVF column (column 3) indicate the number of control intervals (CIs) that are used in DOVF and IOVF respectively.

    If the number of root anchor points in column 1 is significantly less than the total number of root anchor points per UOW, and if fair amount of DOVF or IOVF CIs are used, inefficient randomizing is possibly present in the UOW.

  2. Determine the average length of synonym chains in the UOW.

    The #DB RECORDS column (column 9) indicates the number of database records in the UOW. Divide this number by the number in the #RAPS column (column 1) to determine the average number of database records per used RAP CIs. If this value is larger than what you expected, consider tuning the randomizer.

    Note: The #DB RECORDS column is printed only when the utility that generates this report runs in a specific run mode.
  3. Determine the free space utilization in RAP CIs.

    The efficiency of free space utilization in RAP CIs of the UOW can be determined by analyzing the values in columns 4, 5, 6, and 7.

    The #RAPS USE OVFL CI'S column (column 4) indicates the number of RAP CIs that contain a pointer to a segment data in either DOVF or IOVF. The %RAP FS AVG column (column 5), the %RAP FS MAX column (column 6), and the %RAP FS MIN column (column 7) indicate the average, maximum, and minimum percentages of free space in used RAP CIs. Empty RAP CIs are not included in these values.

    If many RAP CIs use overflow despite the average percentage of free space in RAP CIs is fairly high, it is highly possible that either free space in the RAP CIs is fragmented or fair amount of segment occurrences in the UOW were deleted. In either case, free space reorganization might remedy the overflow from RAP CIs in the UOW.

  4. Determine the free space utilization in DOVF.

    You can determine the efficiency of free space utilization in the DOVF section of the UOW by analyzing the values in columns 3 and 8.

    The %DOVF PART FS column (column 8) indicates the percentage of free space in the DOVF section of the UOW. This value includes all the free space elements that are in the used DOVF CIs and all the empty DOVF CIs.

    If an IOVF CI (column 3) is used by the UOW, despite the fact that the percentages of free space in the DOVF section are fairly high, it is likely that free space in the DOVF section is fragmented or a fair amount of segment occurrences in the DOVF section were deleted. In either case, free space reorganization might resolve the overflow to IOVF.

  5. Determine the length of database records.

    You can determine the average number of database records per used RAP CIs by analyzing the values in columns 1, 9, and 10.

    The #DB RECORDS column (column 9) indicates the number of database records. Divide this number by the NUMBER CI'S USED #RAPS column (column 1) to determine the average number of database records per used RAP CIs.

    Multiply the calculated value with the average length of database records (the RECORD LENGTH STATS - AVERAGE column, column 10). If the resulting value exceeds the CI size of the area, the randomizing to the RAPs in the UOW might not be optimal or the CI size might not match the average database record length.

    If the randomizing is optimal or satisfactory, consider expanding the CI size to improve I/O performance.

    Also, check the maximum record length (column 11) to determine if a large database record exists in the UOW. When large database records exist, your tuning step might require optional steps, which are discussed in the tuning topics in Tuning DEDB areas.

    Note: The #DB RECORDS column (column 9) and the RECORD LENGTH STATS columns (columns 10, 11, and 12) are printed only when a request is made in a utility job.