This report supplies performance data on the CMS (Catalog Management System). Data is collected separately for each catalog. A report is output for each pubset imported during the last monitoring cycle and for all private volumes. The home pubset is marked. Detailed information on CMS is contained in the manuals “HIPLEX MSCF” [8] and “Introduction to System Administration” [6].
Multiprocessor systems
In a multiprocessor system, the CMS report supplies access data for the tasks of the system which manages the catalog (LOCAL) as well as access data for tasks running on remote computers. This data is listed under REMOTE.
In an integrated shared pubset, access by so-called slave sharers is also recorded on the pubset master (SHARED identifier).
Report output
The REPORT CMS statement is used to request the output of the CMS report.
The report can only be output if the SM2 administrator has started the monitoring process using the statement START-MEASUREMENT-PROGRAM TYPE=*CMS.
The SELECT-CMS-PUBSET statement can be used to specify the set of reports to be output. By default, all pubsets and all private disks are output.
HOST0001 SM2 CMS CYCLE: 60 S SAMPLES: 144 <date> <time>
CAT-ID: 2OS6(SLAVE) HIT-RATE: 0 % #-BMT: 0 (PAGEABLE)
-------+--------+---------+---------+-------+-------+-------+--------+---------
REQ. QU| SER. QU| BLOCK QU|USER-ID Q| CE QU | SFP QU| SMP QU| PH-RD/S| PH-WRT/S
-------+--------+---------+---------+-------+-------+-------+--------+---------
0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.0 | 0.0
-------+--------+---------+---------+-------+-------+-------+--------+---------
------------------+-----------+-----------+-----------+-----------+------------
ITEMS | READ | READ-LBN | SCAN | UPD-RENAME| WRITE/CLEAR
------------------+-----------+-----------+-----------+-----------+------------
RESPONSETIME (MS)| | | | |
LOCAL-FILE (/S)| 12.8 | 0.9 | 10.3 | 0.0 | 9.6
JV (/S)| 0.6 | 0.0 | 0.0 | 0.0 | 0.2
REMOTE-FILE (/S)| 0.0 | 0.0 | 0.0 | 0.0 | 0.0
JV (/S)| 0.0 | 0.0 | 0.0 | 0.0 | 0.0
SHARED-FILE (/S)| 0.0 | 0.0 | 0.0 | 0.0 | 0.0
JV (/S)| 0.0 | 0.0 | 0.0 | 0.0 | 0.0
------------------+-----------+-----------+-----------+-----------+------------
Monitoring information
CAT-ID | Catalog identifier |
POOL OF PRIVATE DISKS | This is output for private disks in place of CAT-ID. The monitoring data refers to the F1 label accesses of all private disks. |
HOME | The catalog is marked with home pubset if it belongs to the home pubset. |
MASTER | If the catalog belongs to a shared pubset and the SM2 executes on the associated master processor, the catalog is identified as the MASTER catalog. |
SLAVE | If the catalog belongs to a shared pubset and the SM2 executes on a slave processor, the catalog is identified as the SLAVE catalog. |
HIT-RATE | No read access is required if a wanted catalog entry is already in a CMS storage area. HIT-RATE is the percentage of such (nonphysical) accesses of all read accesses. |
| Number of buffer management tables or buffers used by CMS (refer to the “Introduction to System Administration” manual [6]). |
PAGEABLE | Buffers are pageable (class 4 memory) |
RESIDENT | Buffers are resident (class 3 memory) (refer to the “Introduction to System Administration” manual [6]). |
REQ. QU | Average number of requests waiting for the release of a buffer management table or currently using a buffer management table. If the request queue is always long, this means that there are too few buffers compared to the number of CMS calls, i.e. that the number of buffer management tables should be increased. If the maximum number of buffer management tables is already being used, the pubset can be subdivided into smaller pubsets. The desired number of buffer management tables can be specified at system setup using the BMTNUM parameter. |
SER. QU | Average number of requests waiting for a serialization lock for exclusively searching a suitable buffer management table or currently using this lock (relevant for system engineers) |
BLOCK QU | Average number of requests waiting for an exclusive lock for processing a block of a partition or currently using this lock (a partition corresponds to a user ID) |
USER-ID QU | Average number of requests waiting for a partition lock or currently using this lock. This lock is not normally exclusive. |
CE QU | Average number of requests waiting for an exclusive lock for processing a catalog entry of a block of a partition or currently using this lock |
SFP QU | Average number of requests to the speedcat for single-feature pubsets |
SMP QU | Average number of requests to the catalog index for systemmanaged pubsets |
PH-RD/S | Number of physical read accesses per second |
PH-WRT/S | Number of physical write accesses per second |
The table columns contain the following data: | |
ITEMS | Name of the monitored variable |
READ | Number of read accesses to catalog entries without LBN specification |
READ-LBN | Number of read accesses to catalog entries with LBN specification |
SCAN | Number of SCAN accesses to catalog entries |
UPD-RENAME | Number of write accesses to catalog entries merely involving manipulation of file attributes |
WRITE/CLEAR | Number of write or erase accesses to catalog entries |
The table lines contain the following data: | |
RESPONSETIME (MS) | Global response times for the above data for files and job variables and for all accesses, regardless of their origin (local or remote computer). |
LOCAL-FILE (/S) | File entry accesses originating from the local computer. |
JV (/S) | JV entry accesses originating from the local computer. |
REMOTE-FILE (/S) | File entry accesses originating from a remote computer. |
JV (/S) | JV entry accesses originating from a remote computer. |
SHARED-FILE (/S) | File entry accesses by a slave. |
JV (/S) | Job variable entry accesses by a slave. |