Performance Report columns

• Synchronous DBR
• Synchronous NDBR
• Synchronous Wrt
• Synchronous Sum
• Synchronous Max
• Async Sum
• Async Max

• The percent of Device Cache Read Hit for each arm.
• The percent of Controller Cache Read Hit for each arm.
• The percent of efficiency of write cache

Device read

Device Read is the number of Device Cache Read Hits (DSDCRH) divided by number of Device Read Operations (DSDROP), expressed as a percent

Controller read

Controller Read is the number Controller Cache Read Hits (DSCCRH) divided by number of Read Commands (DSRDS), expressed as a percent.

Write efficiency

Write efficiency is the difference between Write Commands (DSWRTS) and Device Write Operations (DSDWOP) divided by Write Commands (DSWRTS), expressed as a percent.

EACS Read

The percent of read hits by the Extended Adaptive Cache Simulator.

EACS Resp

The percent of response time improvement by the Extended Adaptive Cache Simulator.

0

Number of zero seeks

1/12

Number of seeks between 0 and 1/12 of the disk

1/6

Number of seeks between 1/12 and 1/6 of the disk

1/3

Number of seeks between 1/6 and 1/3 of the disk

2/3

Number of seeks between 1/3 and 2/3 of the disk

>2/3

Number of seeks greater than 2/3 of the disk

MB

Millions of bytes available on the disk.

Percent

Percent of space available on the disk.

• No route was found to transmit the datagrams to their destination.
• Lack of buffer space.

Long Wait

Elapsed times in the state (such as waiting for the next transaction or lock-wait time).

Active/Rsp

During transaction processing, the time the job spends (either waiting or active) while it holds an activity level. At the end of a transaction (on the transaction totals line), this is the time the job spent processing the transaction in an activity level, for long waits caused by locks, and in the ineligible state.

Inel Wait

The time the job spent in the ineligible wait state waiting for an activity level.

Constant

The portion of exceptional wait time held constant as throughput increases.

Variable

The portion of exceptional wait time that varies as throughput increases.

Active Time - [
(multiplier X CPU X Beginning Activity Level) +
(Number of synchronous disk I/O operations X .010)]

• 0=*FIXED, which indicates the system does not dynamically adjust the paging characteristics of the storage pool. The system uses default values.
• 3=*CALC, which indicates the system dynamically adjusts the paging characteristics of the storage pool for optimum performance.

A

Autostart

B

Batch

BD

Batch immediate (Transaction only)

Note: The batch immediate values are shown as BCI on the Work with Active Job display and as BATCHI on the Work with Subsystem Job display.

BE

Batch evoke (Transaction only)

BJ

Batch pre-start job (Transaction only)

C

Programmable workstation application server, which includes 5250 emulation over APPC and iSeries™ Access host servers running either APPC or TCP/IP. A job is reported as a iSeries Access server if any of the following items are true:

• Incoming APPC evoke requests one of the server program names. This also applies to the pre-started jobs for the QSERVER, QCMN, and QSYSWRK subsystems that are already waiting for the named program.
• Incoming IP port number corresponds to one of the service name-description-port-numbers. This also applies to the pre-started jobs for the QSERVER, QCMN, and QSYSWRK subsystems that are already waiting for the assigned IP port number.
• Incoming IPX socket number corresponds to one of the service name-description-port-numbers. This also applies to the pre-started jobs for the QSERVER, QCMN, and QSYSWRK subsystems that are already waiting for the assigned IPX port number.
• Incoming 5250 display emulation jobs that come from APPC data streams sent by 5250 emulation under OS/2^® Communications Manager or WARP equivalent.

D

Target distributed data management (DDM) server

I

Interactive. Interactive includes twinaxial data link control (TDLC), 5250 remote workstation, and 3270 remote workstation. For the Transaction Report, this includes twinaxial data link control (TDLC), 5250 remote workstation, 3270 remote workstation, SNA pass-through, and 5250 Telnet.

L

Licensed Internal Code task

M

Subsystem monitor

P

SNA pass-through and 5250 Telnet pass-through. On the Transaction Report, these jobs appear as I (interactive).

R

Spool reader

S

System

W

Spool writer, which includes the spool write job, and if Advanced Function Printing™ (AFP™) is specified, the print driver job.

WP

Spool print driver (Transaction only)

X

Start system job

Possible job subtype values include the following:

D

Batch immediate job

E

Evoke (communications batch)

J

Pre-start job

P

Print driver job

T

Multiple requester terminal (MRT) (System/36™ environment only)

3

System/36

Noninteractive job types include:

• Autostart
• Batch
• Evoke
• iSeries Access-Bch
• Server
• Spool
• Distributed data management (DDM) server

• Interactive
• Multiple requester terminal (MRT)
• Pass-through
• System/36

If there are significant differences in the values for transaction-related information shown on these reports, do not use the data until you investigate why these differences exist.

AG

Access group

CB

Commit block

CBLK

Commit block

CD

Controller description

CLS

Class

CMD

Command

CTLD

Controller description

CTX

Context

CUD

Control unit description

CUR

Cursor

DEVD

Device description

DS

Data space

DSI

Data space index

DTAARA

Data area

EDTD

Edit description

FILE

File

JOBD

Job description

JOBQ

Job queue

JP

Journal port

JRN

Journal

JRNRCV

Journal receiver

JS

Journal space

LIB

Library

LIND

Line description

LUD

Logical unit description

MBR

Member

MEM

Database file member

MSGF

Message file

MSGQ

Message queue

ND

Network description

OCUR

Database operational cursor

OUTQ

Output queue

PGM

Program

PROG

Program

PRTIMG

Print image

QDAG

Composite piece - access group

QDDS

Composite piece - data space

QDDSI

Composite piece - data space index

QTAG

Temporary - access group

QTDS

Temporary - data space

QTDSI

Temporary - data space index

SBSD

Subsystem description

TBL

Table

• The IP address in the destination field of the IP header was not a valid address to be received at this entity.
• The protocol was unknown or unsupported.
• Not enough buffer space.

The system-wide average utilization does not include data for mirrored arms in measurement intervals for which such intervals are either in resuming or suspended status.

DB Read

The number of database read requests while the job was in that state.

DB Wrt

The number of database write requests while the job was in that state.

NDB Read

The number of nondatabase read requests while the job was in that state.

NDB Wrt

The number of nondatabase write requests while the job was in that state.

Tot

The total number of DB Read, DB Wrt, NDB Read, and NDB Wrt requests.

GETDR

Get direct

GETSQ

Get sequential

GETKY

Get by key

GETM

Get multiple

PUT, PUTM

Add a record

UDR

Update, delete, or release a record

• SDLC
• ASYNC
• BSC
• X25
• TRLAN
• ELAN (Ethernet)
• IDLC
• DDI
• FRLY
• PPP

• W--(Wait state) not holding an activity level.
• A--(Active or wait state) holding an activity level.
• I--(Ineligible state) waiting for an activity level.

The table below shows the possible job state transitions. For example, from W to A is y, or yes, which means it is possible for a job to change from the wait state to the active state.

DB Read

The average number of synchronous database read requests per transaction.

DB Write

The average number of synchronous database write requests per transaction.

NDB Read

The average number of synchronous nondatabase read requests per transaction.

NDB Write

The average number of synchronous nondatabase write requests per transaction.

Sum

The sum of the averages of the synchronous DB READ, DB WRITE, NDB READ, and NDB WRITE requests (the average number of synchronous I/O requests per transaction for the job).

DB Read

The number of synchronous database read requests per transaction.

DB Wrt

The number of synchronous database write requests per transaction.

NDB Read

The number of synchronous nondatabase read requests per transaction.

NDB Wrt

The number of synchronous nondatabase write requests per transaction.

Sum

The sum of the synchronous DB Read, DB Wrt, NDB Read, and NDB Wrt requests (the number of synchronous I/O requests per transaction).

• The system assigns thread IDs sequentially. When a job is started that uses a job structure that was previously active, the thread ID that is assigned to the initial thread is the next number in the sequence.
• The first thread of a job is assigned a number.
• Any additional threads from the same job are assigned a number that is incremented by 1. For example:

  Job Name   User Name/  Job Number
              Thread
  QJVACMDSRV   SMITH       023416
  QJVACMDSRV    00000006   023416
  QJVACMDSRV    00000007   023416
  QJVACMDSRV    00000008   023416

  A thread value greater than 1 does not necessarily mean the job has had that many threads active at the same time. To determine how many threads are currently active for the same job, use the WRKACTJOB, WRKSBSJOB, or WRKUSRJOB commands to find the multiple three-part identifiers with the same job name.

(System, Component) Percentage of available processing unit time used by the partition. For a multiple-processor system, this is the average use across all processors. For dedicated partitions, Total CPU Utilization is replaced by a utilization value for each processor in the partition. Here is an example of this part of the display for a dedicated partition with two processors:

 Average CPU utilization  . . . . . :    41.9
 CPU 1 utilization  . . . . . . . . :    41.7
 CPU 2 utilization  . . . . . . . . :    42.2

In shared processor partitions, individual CPU utilization rows are not printed.

Note: This value is taken from a system counter. Other processing unit uses are taken from the individual job work control blocks (WCBs). These totals may differ slightly. For uncapped partitions, Total CPU utilization might exceed 100 percent.

• No route was found to transmit the datagrams to their destination.
• Lack of buffer space.

A

Autostart

B

Batch

BD

Batch immediate (Transaction only)

Note: The batch immediate values are shown as BCI on the Work with Active Job display and as BATCHI on the Work with Subsystem Job display.

BE

Batch evoke (Transaction only)

BJ

Batch pre-start job (Transaction only)

C

Programmable work station application server, which includes 5250 emulation over APPC and iSeries Access host servers running either APPC or TCP/IP. You can find the host server information under the Host server administration topic in the iSeries Information Center. A job is reported as a iSeries Access server if any of the following items are true:

• Incoming APPC evoke requests one of the server program names. This also applies to the pre-started jobs for the QSERVER, QCMN, and QSYSWRK subsystems that are already waiting for the named program.
• Incoming IP port number corresponds to one of the service name-description-port-numbers. This also applies to the pre-started jobs for the QSERVER, QCMN, and QSYSWRK subsystems that are already waiting for the assigned IP port number.
• Incoming IPX socket number corresponds to one of the service name-description-port-numbers. This also applies to the pre-started jobs for the QSERVER, QCMN, and QSYSWRK subsystems that are already waiting for the assigned IPX port number.
• Incoming 5250 display emulation jobs that come from APPC data streams sent by 5250 emulation under OS/2 Communications Manager or WARP equivalent.

D

Target distributed data management (DDM) server

I

Interactive. For the Component Report, this includes twinaxial data link control (TDLC), 5250 remote workstation, and 3270 remote workstation. For the Transaction Report, this includes twinaxial data link control (TDLC), 5250 remote workstation, 3270 remote workstation, SNA pass-through, and 5250 Telnet.

L

Licensed Internal Code Task

M

Subsystem monitor

P

SNA pass-through and 5250 Telnet pass-through. On the Transaction Report, these jobs appear as I (interactive).

R

Spool reader

S

System

W

Spool writer, which includes the spool write job, and if Advanced Function Printing (AFP) is specified, the print driver job.

WP

Spool print driver (Transaction only)

X

Start the system

The possible job subtypes are:

D

Batch immediate job

E

Evoke (communications batch)

J

Pre-start job

P

Print driver job

T

Multiple requester terminal (MRT) (System/36 environment only)

3

System/36

(System)

The disk type.

(Transaction)

The type and subtype of the job.

(Transaction)

For the Seize/Lock Conflicts by Object section, the type of seize/lock conflict.

• The system assigns thread IDs sequentially. When a job is started that uses a job structure that was previously active, the thread ID that is assigned to the initial thread is the next number in the sequence.
• The first thread of a job is assigned a number.
• Any additional threads from the same job are assigned a number that is incremented by 1. For example:

  Job Name   User Name/  Job Number
              Thread
  QJVACMDSRV   SMITH       023416
  QJVACMDSRV    00000006   023416
  QJVACMDSRV    00000007   023416
  QJVACMDSRV    00000008   023416

  A thread value greater than 1 does not necessarily mean the job has had that many threads active at the same time. To determine how many threads are currently active for the same job, use the WRKACTJOB, WRKSBSJOB, or WRKUSRJOB commands to find the multiple three-part identifiers with the same job name.

EVT

Event Wait. A long wait that occurs when waiting on a message queue.

EOTn

End of transaction for transaction for type n. These codes are in the wait code column, but they are not wait codes. They indicate transaction boundary trace records.

EORn

End of response time for transaction n. These codes are in the wait code column, but they are not wait codes. They indicate transaction boundary trace records.

Error Responses

(Component> The number of responses in error.

HDW

Hold Wait (job suspended or system request).

LKRL

Lock Released. The job released a lock it had on the object named on the next detail line of the report (OBJECT --). The job that was waiting for the object is named on this line (WAITER --) along with the amount of time the job spent waiting for the lock to be released.

LKW

Lock Wait. If there are a number of these, or you see entries with a significant length of time in the ACTIVE/RSP* column, additional analysis is necessary. The LKWT report lines that precede this LKW report line show you what object is being waited on, and who has the object.

LKWT

Lock Conflict Wait. The job is waiting on a lock conflict. The time (*/ time /*) is the duration of the lock conflict and, though not equal to the LKW time, should be very close to it. The holder of the lock is named at the right of the report line (HOLDER --). The object being locked is named on the next report line (OBJECT --).

SOTn

Start of transaction n. These codes are in the wait code column, but they are not wait codes. They indicate transaction boundary trace records.

SWX

Short Wait Extended. The short wait has exceeded a 2-second limit and the system has put the transaction into a long wait. This long wait must be charged to the transaction response time. In other words, this active-to-wait transaction does not reflect a transaction boundary.

SZWG

(Transaction) Listed in the Wait Code column, Seize Wait Granted. The job was waiting on a seize conflict. The original holder released the lock that it had on the object, and the lock was then granted to the waiting job. The job that was waiting for the object is named on this line (WAITER --) along with the amount of time the job spent waiting for the seize conflict to be released. The object that is held is named on the next line of the report (OBJECT --).

SZWT

Seize/Lock Conflict Wait. The job is waiting on a seize/lock conflict. The time (*/ time /*) is the duration of the seize/lock conflict, and is included in the active time that follows it on the report. The holder of the lock is named at the right of the report line (HOLDER --). The object being held is named on the next report line (OBJECT --).

TSE

Time Slice End. The program shown in the stack entry labeled LAST is the program that went to time slice end. Every time a job uses 0.5 seconds of CPU time (0.2 seconds on the faster processors) between long waits, the system checks if there are jobs of equal priority on the CPU queue. If there are, then the next job with equal priority is granted the CPU and the interrupted job is moved to the queue as the last of equal priority. The job, however, retains its activity level. This is an internal time slice end. When a job reaches the external time slice value, there can be a job state transition from active to ineligible if another job is waiting for an activity level. When a job is forced out of its activity level, its pages are liable to be stolen by other jobs, and cause additional I/O when the job regains an activity level. The IBM-supplied default values of 2 seconds for interactive jobs and 5 seconds for batch jobs may often be too high, especially for the high-end processors. As an initial value, set the time slice at 3 times the average CPU seconds per transaction.

WTO

Wait Timed Out. The job has exceeded the wait time-out limit defined for a wait (such as a wait on a lock, a message queue, or a record).

In the Job Trace Analysis I/O Summary, this is the time that the job was inactive, probably due to typing or think time by the user.