Monitoring Directory Search Activity with PMOMON › D3: LNKLST Library Searches – Summary

D3: LNKLST Library Searches – Summary

The D3 display shows statistics on searches of LNKLST library directories. This display does not describe BLDL/FIND requests. For information on the difference between BLDL/FIND requests and library directory searches, see the section D2: BLDL/FIND Request Activity.

The following screen shows a sample D3 display:

PMO 4.4 ( Active ) Sys=IPOX  13NOV10:09:10:50  PMO Active Time=5.500 hrs

           D3    LNKLST Library Searches - Summary of All Sets

                                            % of    Avg Elap   Elap Time
                                    Count   Total     Time      Savings

LNKLST Library Searches        =   294,203   100    0.000931S    2.78H

  Managed List                 =   286,770    97    0.000048S    2.78H
     Member Found              =   116,630    40                 1.13H
     Member Not Found          =   170,140    58                 1.65H

  Library Lookaside (LLA)      =         0     0    0.000000S
     Member Found              =         0     0
     Member Not Found          =         0     0

  I/O Library Searches         =     7,433     3    0.034989S
     Member Found              =       725     0
     Member Not Found          =     6,708     2

Row Headings

The row headings on the D3 display are described in the following:

LNKLST Library Searches

All searches of LNKLST library directories.

Managed List

Searches of LNKLST library directories that the managed list resolved.

Member Found

Searches of LNKLST library directories in which the member was found.

Member Not Found

Searches of LNKLST library directories in which the member was not found. For information on how CA PMO calculates this field, see the discussion of the Count column heading in the following section.

Library Lookaside (LLA)

LNKLST searches that were resolved by Library Lookaside (LLA).

An asterisk (*) in this field indicates that you excluded one or more LNKLST libraries from the managed list. If you see an asterisk, it may represent LNKLST libraries that you did not code on the INCLLNK parameter.

Otherwise, the asterisk indicates that you specified the EXCLLNK parameter or the F PMO,EXCLLNK command to exclude one or more libraries from CA PMO management. The D4 display shows the names of the excluded libraries.

I/O Library Searches

LNKLST searches resolved by DASD I/O.

An asterisk (*) in this field indicates that you specified the EXCLLNK parameter or the F PMO,EXCLLNK command to exclude one or more libraries from the managed list. The D4 display shows the names of the excluded libraries.

Note: When you exclude LNKLST libraries, you reduce the percentage of LNKLST searches that CA PMO can resolve.

Column Headings

The column headings on the D3 display are described in the following sections.

Count

There are differences in the way CA PMO calculates the statistics depending on whether LLA is active on the system or not. If LLA is not active, CA PMO counts a member as found or not found by the following method:

• CA PMO adds one to the Member Found count for the first library in the concatenation that contains the member.
• CA PMO adds one to the Member Not Found count for each library in the concatenation preceding the one that contains the member.
• If CA PMO does not find the member, it adds one to the Member Not Found count for each library in the concatenation.
• If CA PMO does not find the member, but it does find an entry created with the parameter RETNF=YES, it adds one to the Member Not Found count of the managed list for each library in the concatenation. This eliminates DASD I/O for members that do not exist.

The following table shows how CA PMO breaks down a BLDL/FIND request against the following LNKLST concatenation:

LNKLST library 1= SYS1.LIB1
LNKLST library 2= SYS1.LIB2
LNKLST library 3= SYS1.LIB3

CA PMO searches a concatenation in the order in which the libraries are listed. For every library directory that CA PMO searches that does not contain the member, it adds one to the Member Not Found count. When CA PMO finds the member, it adds one to the Member Found count and the search ends.

If LLA is active, CA PMO counts the members as found or not found by the following method:

• CA PMO adds one to the Member Found count for the library that contains the member.
• If CA PMO does not find the member, both counts for individual libraries in the concatenation and counts for summary LNKLST searches remain unaffected.
• If CA PMO does not find the member, but it does find an entry created because of the parameter RETNF=YES, it adds one to the Member Not Found count of the managed list for summary LNKLST searches. This eliminates LLA searches for members that do not exist. Counts for individual libraries in the concatenation remain unaffected.
• The reason for this differentiation is that if the search for a member is made by DASD I/O, then the searches need to be done to all libraries in the concatenation prior to the library that contains the member. However, if the search for a member is done by LLA, all libraries in the concatenation are searched at once by a single search. The Member Not Found counts for individual libraries would have no meaning in this case. You should recycle CA PMO each time LLA is started or stopped to prevent invalidating the statistics caused by the differences in method of calculating Member Found and Member Not Found counts.

% Of Total

A breakdown of LNKLST library directory searches into the percentage that CA PMO handled and the percentage it did not handle.

Avg Elap Time

The average time it takes to resolve a LNKLST library directory search:

• For searches of the managed list, CA PMO calculates the average elapsed time during the searches.
• For LLA and DASD I/O searches, CA PMO calculates the average elapsed time during a directory search sampling period that takes place immediately before CA PMO begins managing a library.

CA PMO may either estimate averaged elapsed time or actually measure it:

• For searches of the managed list, CA PMO either divides the time of a particular search among the libraries that would be searched by DASD I/O, or, if LLA is active, the time of a search is credited to the library that contains the member.

  Note: There is one other reason why LLA average elapsed times may be shorter than managed list times. If the more frequently requested members belong to libraries early in the LNKLST concatenation, and the less frequently requested ones are from later libraries, the managed list search times may be divided among fewer libraries. This is because the less frequently requested members may not be in the managed list.

  Note: CA PMO uses this type of estimate because the managed list and LLA are searched once, at most, as a result of a BLDL/FIND request.

  Note: The D3 display may show that LLA has a shorter average elapsed time than the managed list. There may be several reasons for this. If you run LLA with CA PMO, each LLA refresh increases the elapsed times for the managed list. After an LLA refresh, CA PMO must remove all LNKLST members from the managed list, and the first BLDL/FIND request for a member after the refresh must be resolved by LLA. (CA PMOplaces a new entry for the member at the bottom of the managed list after the first BLDL/FIND request for it.) You should discontinue use of LLA if it is refreshed often. Also, if CA PMO was started during a non-busy period on your system, the average elapsed time for LLA searches, measured during startup, may not reflect the paging contention during busier periods when CA PMO is helping the most.

There is one other reason why LLA average elapsed times may be shorter than managed list times. If the more frequently requested members belong to libraries early in the LNKLST concatenation, and the less frequently requested ones are from later libraries, the managed list search times may be divided among fewer libraries. This is because the less frequently requested members may not be in the managed list.

For I/O searches, CA PMO actually measures the elapsed time for each library search:

• If there were no I/O searches since CA PMO became active or since the last F PMO,RESET command, CA PMO uses a simple average of the time for each library search during the directory search sampling period. This figure has no bearing on the time shown in the LNKLST Library Searches row.
• If there were I/O searches, CA PMO calculates a weighted average for average elapsed time for I/O. CA PMO weights the simple average obtained during the directory search sampling period according to the number of DASD searches for each library while CA PMO is active.

Elap Time Savings

The elapsed time that CA PMO saved on searches of LNKLST library directories. This figure is the difference between the average elapsed time for managed list searches and the average elapsed time for one of the following:

• LLA searches, if LLA is running for the entire time that CA PMO is active
• I/O searches, if LLA is inactive while CA PMO is active

If you see three asterisks (***), CA PMO did not save any elapsed time.