FCX102, System Counters – SYSTEM

Performance Toolkit Subcommand: SYSTEM.

The System Counters report (FCX102):
Figure 1. Layout of System Counters Report (FCX102)
  FCX102      CPU nnnn  SER nnnnn  Interval HH:MM:SS - HH:MM:SS    Perf. Monitor

  Operation                Count Rate/s    Operation                Count Rate/s
  Real SSCH instructions    9673    161    Real CSCH instructions       1     .0
  Real HSCH instructions       0     .0    El. time slice drops       336    5.6
  SVC instr. simulated         0     .0    SVC interrupts reflectd      0     .0
  SVC 76 reflected             0     .0    Diagnose I/O requests      489    8.2
  FP external call simul.      0     .0    FP partial executions     7449    124
  Fast-path SIGP simulat.      0     .0    FP simul of Diag.X'44'       0     .0
  Diag.X'9C', Not Ready       18     .3    Diag.X'9C', Wait State     642   10.7
  Diag.X'9C', Soft Wait        7     .1    Diag.X'9C', Alr.Dispat.   9553    159 
  FP successful x-lates       19     .3    CCW chains not FP-elig.      0     .0
  Fast-path aborts             0     .0    Total FP xlate attempts     19     .3
  Nr. of SIE executions    55674    927    Nr. of SIE intercepts    30232    503
  Entries to enabled wait  16548    275    CP issued Diag.X'44'         0     .0 
  HPF write                99999   99.9    HPF done                 99999   99.9 

  Storage Management
  Subpool FREE req bk <2G  34118    569    Total FREE req bk <2G    34118    569
  Subpool FREE req bk >2G 178404   2973    Total FREE req bk >2G   178407   2973
  Reserved                                 Storage fast clears        294    4.9
  Avail lst frame req <2G   1052   17.5    Avail lst frame req >2G    740   12.3
  Available lst empty <2G      0     .0    Available lst empty >2G      0     .0
  Demand scan 1st pass       ---    ---    Demand scan 2nd pass       ---    ---
  Demand scan emergency      ---    ---    Demand scan not satisf.    ---    ---
  Dispatch lst pgs stolen    ---    ---    Shared stor. pgs taken     ---    ---
  Pgs from dormant users     ---    ---    Eligible lst pgs stolen    ---    --- 
  Fast PGINs from XSTORE       0     .0    Pages taken for FREE         0     .0
  PGOUTs main to XSTORE        0     .0    Slow PGINs from XSTORE       0     .0
  XSTORE allocations           0     .0    Reserved
  Glbl cycl list searched      0     .0    XSTORE releases              0     .0
  Migr thresh buf increas      0     .0    Migr. target time reset      0     .0
  Page migr. from dormant      0     .0    Migr thresh buf lowered      0     .0
  Page migr. from active       0     .0    Dormant with page migr.      0     .0
  Shared pages migrated        0     .0    Active with page migrat      0     .0
  Blocks migrated from CP      0     .0    Shared sys with pg migr      0     .0
  Blocks migrated to DASD      0     .0    PGMBKs sel. during migr      0     .0
  No I/O for pg migration      0     .0    XSTORE migr invocations      0     .0
  Pg not referenced (STL)      0     .0    Pg not referenced (MIG)      0     .0
  Single system pg reads       0     .0    Page blocks read             0     .0
  Pages read from DASD         0     .0    Single guest page reads      0     .0
  Spool pages read             0     .0    Pages written to DASD        0     .0
  Total pgs to/from DASD       0     .0    Spool pages written          0     .0
  
  Minidisk Cache Management
  MDC page reads               9     .2    MDC page writes              0     .0
  MDC wait queue inserts       0     .0    MDC fair share exceeded      0     .0
  Compl. successful reads    412    6.9    MDC blocks invalidated       0     .0
  Total MDC read requests    412    6.9    Total invalidate req.        0     .0

  Memory Constraint Relief      
  Redrive attmps anywhere      0     .0    Task redrives anywhere       0     .0 
  Redrive attempts <2G         0     .0    Task redrives <2G            0     .0 
  Redrive attempts >2G         0     .0    Task redrives >2G            0     .0  

  Command ===> _
  F1=Help  F4=Top  F5=Bot  F7=Bkwd  F8=Fwd  F12=Return

The SYSTEM report provides additional detailed information about the system's operation, especially in the areas of storage management and minidisk caching. It is intended to allow a more detailed analysis of information shown on other reports (CPU and STORAGE, for example), if you suspect a problem in one of these areas.

Notes:
  1. When running Performance Toolkit against z/VM 5.3 or earlier data, you will see dashes in all fields on this report.
  2. When it is running against z/VM 5.4 data, Performance Toolkit can use trend files created using monitor data from any release; in which case, older data might cause dashes to appear in some fields. Note, however, that Performance Toolkit cannot use trend files that were actually created using older data for this report.

Description of SYSTEM Display Fields:

Operation

This section describes the events that are monitored.

Count
The count in the selected interval, that is, during the last measurement interval or since the last RESET, for "current" or "average" data selection, respectively.
Rate/s
The number of occurrences per second in the selected interval.
Events Shown:
Real SSCH operations
The number of real SSCH operations (Start Subchannel) run by the system.
Real CSCH instructions
The number of real CSCH instructions (Clear Subchannel) run by the system.
Real HSCH instructions
The number of real HSCH instructions (Halt Subchannel) run by the system.
El. time slice drops
The number of elapsed time slice drops in the interval. A virtual machine's (or more correctly: a VMDBK's) stay in the dispatch list ends when it has used up its allocated time (the 'elapsed time slice'), or when the virtual machine's work has ended and it becomes idle.
SVC instr. simulated
The number of SVC instructions simulated by the system for guest operating systems.
SVC interrupts reflectd
The number of SVC interrupts reflected to guest systems.
SVC 76 reflected
The number of SVC 76 reflected to guest systems.
Diagnose I/O requests
The number of virtual I/O requests via Diagnose codes X'18', X'20', X'58', X'A4', and X'A8'. Other virtual I/O activity (for example, via SIO, SSCH) is not included in this count. See FCX100, CPU Load and Transactions – CPU, and FCX101/FCX181 Remote Performance Log – REDISP, REDHIST for the total virtual I/O activity.
FP external call simul.
The number of fast path reflections of guest external call interrupts.
FP partial executions
The number of fast path partial execution intercepts.
Fast-path SIGP simulat.
The number of fast path simulations of SIGP external call instructions.
FP simul. of Diag.X'44'
The number of fast path simulations of Diagnose X'44'.
Diag.X'9C', Not Ready
Count of times a guest issued Diagnose X'9C' and the target was in a VMDSTATE < VMDREADY.
Diag.X'9C', Wait State
Count of times a guest issued Diagnose X'9C' and the target was found in wait state.
Diag.X'9C', Soft Wait
Count of times a guest issued Diagnose X'9C' and the issuer had to soft-wait on the target.
Diag.X'9C', Alr.Dispat.
Count of times a guest issued Diagnose X'9C' and the target was already dispatched.
FP successful x-lates
The number of successful fast-path CCW translations.
CCW chains not FP-elig.
The number of times a CCW chain was found to be not eligible for fast-path translation.
Fast-path aborts
The number of times a fast-path translate attempt was aborted.
Total FP xlate attempts
The total number of fast-path CCW translate attempts.
Nr. of SIE executions
The number of times the SIE operation (Start Interpretive Execution) was run.
Nr. of SIE intercepts
The number of times an SIE operation (Start Interpretive Execution) ended due to an intercept (privileged operation intercepted for handling by CP).
Entries to enabled wait
The number of guest machine entries to an enabled wait state.
CP issued Diag.X'44'
Count of Diagnose X'44's issued by CP.
HPF Write
Number of times a write channel program for a DASD device was presented for HPF DCW translation.
HPF Done
Number of times HPF DCW translation processing completed successfully for a DASD device.

Storage Management Fields

This block provides many fields with information on central and expanded storage management.
Subpool FREE req bk <2G
The number of requests for free storage which were satisfied from subpools below 2G.
Total FREE req bk <2G
The total number of requests for free storage backed below 2G (sysperm, guestperm, and regular subpooled).
Subpool FREE req bk >2G
The number of requests for free storage which were satisfied from subpools above 2G.
Total FREE req bk >2G
The total number of requests for free storage backed above 2G (sysperm, guestperm, and regular subpooled).
Storage fast clears
The number of fast path clears of real storage (requests for a new page which could be satisfied by a zeroed page without disk I/O).
Avail lst frame req <2G
The number of times that a new frame was requested from the <2G available list.
Avail lst frame req >2G
The number of times that a new frame was requested from the >2G available list.
Available lst empty <2G
The number of times a <2G frame request could not be satisfied from the <2G available list because it was empty. This number should be very low in relation to the total number of <2G frame requests – the available list replenishment task ("demand scan") will usually ensure that a sufficient number of page frames are made available, well before the frames are actually needed.
Available lst empty >2G
The number of times a >2G frame request could not be satisfied from the >2G available list because it was empty. This number should be very low in relation to the total number of >2G frame requests – the available list replenishment task ("demand scan") will usually ensure that a sufficient number of page frames are made available, well before the frames are actually needed.
Demand scan 1st pass
The number of times the available list replenishment task completed after the first pass.

This routine must make a certain number of real storage page frames free. During the first pass, it will take only pages from dormant (that is, inactive) users and unreferenced pages from active users, which should have a minimal impact on system performance.

When running Performance Toolkit against z/VM 6.3 or later data, you will see dashes in this field.

Demand scan 2nd pass
The number of times the available list replenishment task completed after the second pass.

During the second pass of demand scan, referenced pages from active users (in eligible or dispatch list) will also be taken. Since these pages may be needed again soon, the frequency of occurrence should be low. High numbers indicate insufficient real storage for the current workload.

When running Performance Toolkit against z/VM 6.3 or later data, you will see dashes in this field.

Demand scan emergency
The number of times the available list replenishment task completed after the emergency pass.

During the emergency pass of demand scan, referenced shared storage, system pages, and even reserved pages from active users (in eligible or dispatch lists) are taken. These pages have to be taken to keep the system running, but they were probably frequently referenced and will have to be paged in again soon (thrashing). High emergency pass numbers indicate a bad real storage bottleneck.

When running Performance Toolkit against z/VM 6.3 or later data, you will see dashes in this field.

Demand scan not satisf.
The number of times the available list replenishment task ran without satisfying the demand for frames.

This indicates an even more severe storage problem than the 'emergency pass' counts.

When running Performance Toolkit against z/VM 6.3 or later data, you will see dashes in this field.

Dispatch lst pgs stolen
The number of pages taken from users in the dispatch list (unreferenced and referenced pages).

When running Performance Toolkit against z/VM 6.3 or later data, you will see dashes in this field.

Shared stor. pgs taken
The number of shared segment pages taken by the available list replenishment function. SYSTEM and shared segment pages are taken only during the emergency pass of the demand scan, that is, the counts should be low for systems with sufficient real storage.

When running Performance Toolkit against z/VM 6.3 or later data, you will see dashes in this field.

Pgs from dormant users
The number of pages taken from dormant users.

When running Performance Toolkit against z/VM 6.3 or later data, you will see dashes in this field.

Eligible lst pgs stolen
The number of pages taken from users in the eligible list (unreferenced and referenced pages).

When running Performance Toolkit against z/VM 6.3 or later data, you will see dashes in this field.

Pages taken for FREE
The number of pages taken from users by frame table scan.
Fast PGINs from XSTORE
The number of fast path PAGE IN operations (page moves from expanded storage to main storage).
Slow PGINs from XSTORE
The number of PAGE IN operations that did not take the fast path.
PGOUTs main to XSTORE
The number of pages moved out from main storage to expanded storage.
XSTORE allocations
The number of allocations of expanded storage frames.
XSTORE releases
The number of times that an XSTORE frame was released.
Glbl cycl list searched
The number of times the migration task has searched through the global cyclic list of logged-on users.

The MIGRATE task searches through the cyclic list in order to find XSTORE blocks to migrate out and thus free up XSTORE. It chooses blocks to migrate out based on the migrate target time (initially set equal to the XSTORE average age). If half the cyclic list is searched without finding enough migratable blocks, the target time is lowered and an additional search is made through the cyclic list (see also next value).

Migr. target time reset
The number of times the migration target time was reset. See also description of previous value.
Migr thresh buf increas
The number of times the migration threshold buffer was increased.

The migration threshold buffer is the buffer used for block paging migration blocks. Its size is initially set equal to the number of paging devices times a multiplication factor of 10. If the low threshold is raised, the multiplication factor is incremented by 1 and thus the size of the threshold buffer increased.

This is an indication of migration not being able to keep up with XSTORE demand. A high number may indicate concern.

Migr thresh buf lowered
The number of times that the migration threshold buffer was lowered.

This is an indication of migration working well.

Page migr. from dormant
The number of pages migrated from dormant users.
Dormant with page migr.
The number of dormant users which had pages migrated from them.
Page migr. from active
The number of pages migrated from active users.
Active with page migr.
The number of active users which had pages migrated from them.
Shared pages migrated
The number of pages migrated from shared systems.
Shared sys with pg migr
The number of shared systems which had pages migrated from them.
Blocks migrated from CP
The number of blocks migrated from CP (system owned pages).
PGMBKs sel. during migr
The number of PGMBKs selected while migrating. This is an indication of the efficiency of the migration task: the more pages can be migrated from a single PGMBK, the more efficient page migration works, that is, you should see this number in relation to the number of pages migrated.
Blocks migrated to DASD
The number of pages migrated to DASD (the total of the pages migrated from dormant and active users, shared systems, and CP)
XSTORE migr invocations
The number of times the migration task was invoked.
No I/O for pg migration
The number of times no I/O was necessary for migrating a page. If a page already resides on disk, and if it was not referenced since being read for the last time, it can simply be released.
Pg not referenced (MIG)
The number of pages found by the migrate task which have been read in but were not referenced before the migrate task selected them for migration. Block paging may read in many additional pages when a page fault occurs. The ones not used are counted here, that is, a low number is good.
Pg not referenced (STL)
The number of pages found by the steal function which have been read in but were not referenced before the steal function moved them out of main memory to either XSTORE or DASD.
Page blocks read
The number of pages read from DASD (block paging).
Single system pg reads
The number of single-page reads for CP.
Single guest page reads
The number of single-page reads for virtual machines.
Pages read from DASD
The number of pages read from paging DASD. This value includes read operations in progress.
Pages written to DASD
The number of pages written to paging DASD. This value includes write operations in progress.
Spool pages read
The number of pages read from SPOOL. This value includes read operations in progress.
Spool pages written
The number of pages written to SPOOL. This value includes write operations in progress.
Total pgs to/from DASD
The count of pages serviced (reads and writes) for paging and spool. This value is updated after the I/O operation completes. When page or spool I/O is outstanding, its value will be less than the sum of the preceding four counters (pages read from and written to paging DASD and SPOOL).

Minidisk Cache Management Fields

This block provides detailed information on the management of the minidisk cache.
MDC page reads
The number of CMS blocks moved from XSTORE to main storage.

Note that this is not the number of CMS read operations: read misses are also read operations, but they will not result in a MDC page being moved to main storage. Calculate the number of read operations by adding the number of read hits, partial read hits, and read misses (you can find this value on the general STORAGE display).

MDC page writes
The number of minidisk data blocks moved from main storage to XSTORE. This includes blocks which are written to XSTORE as the result of a WRITE operation and blocks which are written to XSTORE after a READ operation for a block which could not be found in expanded storage (read misses).
MDC wait queue inserts
The number of insertions into the waiting queue. The waiting queue keeps track of users that request data which are already on the way to the cache and so do not require another I/O operation, even though the blocks are not yet available in the minidisk cache.
MDC fair share exceeded
The number of times minidisk data blocks were not put into the cache because the fair share limit was exceeded.
Compl. successful reads
The number of MDC reads which were completely successful (all blocks found).
MDC blocks invalidated
The number of MDC blocks which were invalidated.
Total MDC read requests
The number of read requests to MDC.
Total invalidate req.
The total number of invalidate requests.

Memory Constraint Relief delta counters

This block provides detailed information on the Memory Constraint Relief fields.
Redrive attmps anywhere
Count of times HCPPTGRD attempted to redrive tasks waiting for an available frame that can be located anywhere.
Task redrives anywhere
Count of individual task redrives performed by HCPPTGRD for an available frame that can be located anywhere.
Redrive attempts <2G
Count of times HCPPTGRD attempted to redrive tasks waiting for an available frame that can be located only <2G.
Task redrives <2G
Count of individual task redrives performed by HCPPTGRD for an available frame that can be located only <2G.
Redrive attempts >2G
Count of times HCPPTGRD attempted to redrive tasks waiting for an available frame that can be located only >2G.
Task redrives >2G
Count of individual task redrives performed by HCPPTGRD for an available frame that can be located only >2G.

Effect of AVERAGE and INTERIM subcommands:

Overall period totals since last RESET will be shown for all Count fields; overall averages will be displayed for the Rate/s fields.