Analyzing the randomizing routine

Analyze the Synonym Chain Analysis report to understand how efficiently the root segments are distributed by the randomizing routine (randomizer).

Before you begin

Ensure that you can access the Synonym Chain Analysis report for the subject DEDB area.

Not all utilities of IMS Fast Path Solution Pack can produce this report. Use the appropriate utilities and runtime options to generate these reports. For details about utility options and reports that can be generated, see the topic "Generating the analysis reports" in the IMS Fast Path Solution Pack: IMS High Performance Fast Path Utilities User's Guide.

Before you start, consider reviewing the database record profile to obtain the average length of database records, the average number of root I/Os, and the average number of database record I/Os even if you have not received exceptions that are related to the average database record length, root I/O, or record I/O (see Analyzing database record profile). These figures are useful indicators to complete the steps in this topic.

About this task

One of the most frequently overlooked aspects of the database design and implementation process is the selection of the randomizer that best matches the key values of the database records. Using a randomizer that does not match the key values can significantly affect the performance characteristics of a database.

In this task, you will first check the synonym chain distribution, then check the synonym chain effect on root I/Os.

By completing this task, you will be able to evaluate whether the current randomizing logic matches the current distribution of database record key values. If the logic does not match, you will be able to evaluate how this mismatch affects the efficiency of root I/Os.

Procedure

Complete the following steps to determine the randomizing efficiency:

  1. Check the synonym chain distribution.

    Browse the Roots per RAP Distribution Graph in the Synonym Chain Analysis report. The following figure shows an example of the Roots per RAP Distribution Graph in the Synonym Chain Analysis report.

    Figure 1. Roots per RAP Distribution Graph of Synonym Chain Analysis report
    ROOTS PER RAP DISTRIBUTION GRAPH                                      
                                                                          
                                                                1         
                   1    2    3    4    5    6    7    8    9    0         
                   0    0    0    0    0    0    0    0    0    0   % RAPS
               --------------------------------------------------         
       #    0-|*   (234)                                                  
            1-|***   (963)                                                
       R    2-|*******   (2009)                                           
       O    3-|*********   (2834)                                         
       O    4-|**********   (2954)                                        
       T    5-|********   (2493)                                          
       S    6-|******   (1812)                                            
            7-|***   (995)                                                
       /    8-|**   (569)                                                 
            9-|*   (258)                                                  
       R   10-|*   (124)                                                  
       A   11-|*   (32)                                                   
       P   12-|*   (16)                                                   
           13-|*   (7)                                                    
           14-|                                                           
                                                                          
                                                                          
             SYNONYM CHAIN STATISTICS                                     
                                                                          
                 SYN CHAIN LONGEST:      13    RAP RBA: 006C4000          
                 SYN CHAIN MAX I/O:       2    RAP RBA: 006C4000          
                                                                          
                 SYN CHAIN LENGTH :   AVG:    4.49                        
                 NO. ROOTS PER RAP:   AVG:    4.27                    

    Analyze this graph and the synonym chain statistics below the graph to determine the following information:

    • Whether the general distribution of RAP synonym chain lengths is acceptable.
      • In each row, the number on the left represents the number of root segments that are assigned to a RAP. The number of asterisks (*) represents the percentage of the RAPs to which the same number of root segments are assigned. The numbers on the right (in parentheses) represent the number of RAPs. For example, the third row (where the number on the left is 2) represents the percentage and the number of the RAPs to which two root segments are assigned.
      • The average length of RAP synonym chains is shown on the SYN CHAIN LENGTH line in the Synonym Chain Statistics section.
      Generally, if the peak in the graph is near the bottom, the graph suggests that many long synonym chains exist.
    • Whether an exceptionally long synonym chain exists.

      Existence of such synonym chain might indicate that the randomizer does not match the distribution of root sequence key values.

    • Whether the percentage of unused RAPs, which is indicated by the first row in the graph, is high.

      If this number is high and the value for the average length of RAP synonym chains is also high, the current randomizer might not match the distribution of root sequence key values.

  2. Check the synonym chain effect on root I/Os.

    Browse the synonym chain analysis section in the Synonym Chain Analysis report.

    This section of the report shows statistics about the placement of root segments in RAA BASE, DOVF, and IOVF sections of the area. The report also shows the effect on the number of physical I/Os that would be required to read a root segment in a synonym chain for each synonym chain length.

    The first three columns are tabular representations of the Roots per RAP Distribution Graph.

    The next six columns indicate, for each RAP synonym chain length, the following statistics about the root segment occurrences in each section of the area:
    • The average number (AVG.)
    • The maximum number (MAX)
    • The minimum number (MIN)
    • The standard deviation (SDEV)
    The first two columns are for the RAA BASE section, the next two columns are for the DOVF section, and the last two columns are for the IOVF section.
    AVG./SDEV
    The AVG./SDEV columns (the first column for each section) indicate the average number of root segment occurrences over all the synonym chains of that chain length are displayed with the standard deviation from the average.
    MAX/MIN
    The MAX/MIN columns (the second column for each section) indicate the maximum and the minimum number of root segment occurrences among all the synonym chains of that chain length are displayed.
    For each section, a low standard deviation indicates that, within a group of synonym chains that have the same length, the number of root segments that occur in a synonym chain of that group and that reside in that section tends to be very close to the average number, over all synonym chains in that group, of root segment occurrences in that section. A high standard deviation indicates that the root segment occurrences are spread out over a large range of values.
    The last four columns indicate the statistics for physical I/Os as follows:
    AVG.
    The AVG. column indicates the average number of I/Os that are required to read a root segment in the synonym chain of that chain length. The average is calculated based on all the synonym chains that have the same length.
    SDEV
    The SDEV column indicates the standard deviation from the average number.
    MAX and MIN
    The MAX and the MIN columns indicate the maximum and the minimum number of root I/Os among all the synonym chains that have the same length.
    In this section, pay attention to the existence of synonym chains that make the average number of I/Os high or that have an exceptionally high (maximum) number of I/Os.

    The following figure shows the synonym chain analysis section in the Synonym Chain Analysis report.

    Figure 2. Synonym chain analysis section of the Synonym Chain Analysis report
    SYNONYM CHAIN ANALYSIS                                                                  
                                                                                                                 
                                    -----------ROOT PLACEMENT STATISTICS------------    ---PHYSICAL I/O STATS----
                                    --IN RAABASE--   ---IN DOVF----   ---IN IOVF----                             
       #ROOTS/   NO. RAPS     P/C    AVG./    MAX/    AVG./    MAX/    AVG./    MAX/      AVG.    SDEV   MAX  MIN
         RAP                         SDEV     MIN     SDEV     MIN     SDEV     MIN                              
                                                                                                                 
          0           234     1.5      N/A     N/A      N/A     N/A      N/A     N/A       N/A     N/A   N/A  N/A
                                                                                                                 
          1           963     6.3     1.00       1     0.00             0.00              1.00    0.00     1    1
                                                 1                                                               
          2         2,009    13.1     2.00       2     0.00             0.00              1.00    0.00     1    1
                                      0.00       2                                                               
          3         2,834    18.5     3.00       3     0.00             0.00              1.00    0.00     1    1
                                      0.00       3                                                               
          4         2,954    19.3     4.00       4     0.00             0.00              1.00    0.00     1    1
                                      0.00       4                                                               
          5         2,493    16.3     5.00       5     0.00             0.00              1.00    0.00     1    1
                                      0.00       5                                                               
          6         1,812    11.8     6.00       6     0.00             0.00              1.00    0.00     1    1
                                      0.00       6                                                               
          7           995     6.5     7.00       7     0.00             0.00              1.00    0.00     1    1
                                      0.00       7                                                               
          8           569     3.7     8.00       8     0.00             0.00              1.00    0.00     1    1
                                      0.00       8                                                               
          9           258     1.7     9.00       9     0.00             0.00              1.00    0.00     1    1
                                      0.00       9                                                               
         10           124     0.8    10.00      10     0.00             0.00              1.00    0.00     1    1
                                      0.00      10                                                               
         11            32     0.2    11.00      11     0.00             0.00              1.00    0.00     1    1
                                      0.00      11                                                               
         12            16     0.1    12.00      12     0.00             0.00              1.00    0.00     1    1
                                      0.00      12                                                               
         13             7     0.0    12.00      12     1.00       1     0.00              2.00    0.00     2    2
                                      0.00      12     0.00       1                                              
         14             0                                                                                        

What to do next

After you identify the cause of the exceptions, the next step is to determine the actions to tune the subject area. The actions might include tuning of the randomizer to make the database fit the current characteristics of database records and application usage. See Determining actions to resolve exceptions to determine the appropriate tuning tasks.