Important: Do not confuse the term performance tools with the licensed product 5770-PT1 Performance Tools.

Requirement: To take advantage of all the Performance Tools enhancements that are described in this chapter, the system must have the latest levels of PTFs installed.

Operating system functionality: All of the functions that allow the configuration of data collection, to start and end data collection, and to manage the collection objects, are part of the operating system.

Collection Services Enhancements	Changed File	7.1 PTFs
SR-IOV Ethernet physical ports	QAPMETHP	TR6 and SI47870
Job level SQL metrics	QAPMJOBMI	SI47594
IBM Storwize V7000	QAPMDISK	TR6
Active memory sharing deduplication	QAPMSHRMP	SI45848
DMPMEMINF updated to indicate page size		SI44192
SQL full open count	QAPMJOBOS	SI44183
Ethernet link aggregation	QAPMETH	SI43661
12X and P7IOC internal bus data	QAPMBUS QAPMBUSINT	SI43661
Workload groups	QAPMSYSWLC QAPMJOBMI	SI39804
TLBIE metrics	QAPMSYSINT	MF56871 and SI49418

Column name	Description
INTNUM	Interval number: The nth sample database interval that is based on the start time in the Create Performance Data (CRTPFRDTA) command.
DATETIME	Interval date (mmddyy) and time (hhmmss): The date and time of the sample interval.
INTSEC	Elapsed interval seconds: The number of seconds since the last sample interval.
BUIOPB	System bus number: Bus numbering begins with one. Before V5R4, bus numbering began with zero.
BUOPSR	Number of OPSTARTs received: RRCB in server storage.
BUSGLR	Signals received.
BUOPSS	Number of OPSTARTs sent.
BUSGLS	Signals sent.
BURSTQ	Restart queues that are sent.
BUBNAR	Occurrences of BNA received.
BUTPKT	Total packets (sent or received).
BUKBYO	Kilobytes DMAed out.
BUKBYI	Kilobytes DMAed in.
BUNOSR	Normal flow OPSTARTs received.
BUNRDR	Reserved.
BUORQS	OPSTART requests sent.
BUTIMO	Reserved.
BUBNAS	Reserved.
BUQSAS	Queue space available sent.
BUTYPE	Bus type. Supported values are: •‘S’ = SPD bus (no longer supported). •‘P’ = PCI bus. •‘V’ = Virtual bus.
BUCAT	Bus category. This field indicates if this bus record has some special characteristics, which might require a special interpretation of its performance data. Each bit in this field has an independent meaning: •X'00' = No special category applies. •X'01' = This bus is attached to an I/O hub. (Note: The following might be used depending on what happens with switches.) •X’02’= This record represents a switch. The data that is reported is the sum of all buses under the switch. The bus number that is reported is the first bus under the switch.
BUHUB	Hub number. If this bus is associated with an I/O hub, this number is the number of that hub. (Note: An I/O hub can be embedded in the backplane.)
BUMAXRATE	Maximum byte rate. When available from hardware, this rate is the estimated maximum rate that data might be both sent and received in bytes per second through the hardware port.
BUCBSND	Command bytes sent. When available from hardware, this number is the number of command bytes sent through the hardware port.
BUDBSBD	Data bytes sent. When available from hardware, this number is the number of data bytes sent through the hardware port.
BUCBRCV	Command bytes received. When available from hardware, this number is the number of command bytes received through the hardware port.
BUDBRCV	Data bytes received. When available from hardware, this number is the number of data bytes received through the hardware port.

Column name	Description
BUNBR	The hardware assigned number that is associated with the bus or hub.
BUTYPE	Bus type. The supported bus types are: •4 - 12X loop. •6 - I/O hub (might be embedded in the backplane).
INTNUM	Interval number: The nth sample database interval based on the start time specified in the Create Performance Data (CRTPFRDTA) command.
DTETIM	Interval date and time. The date and time of the sample interval.
INTSEC	Elapsed interval seconds: The number of seconds since the last sample interval.
BUNBR	Bus number. The hardware assigned number that is associated with the bus or hub.
BUTYPE	Bus type. Supported bus types are: •4 - 12X Loop •6 - I/O hub (might be embedded in the backplane)
BUDFMT	Bus data format. This field is provided to help you understand what data are instrumented by the hardware components of the bus if there are future differences.
BUATTR1	Bus attribute 1. The meaning of this field depends on the bus type. One row is present for each bus type (BUTYPE) field: •Type 4: Port identifier. One record is present for each supported port.  – 0 = even port  – 1 = odd port •Type 6: Category.  – 0 = Topside port
BUPKTSND	Packets sent. Not supported for type 6.
BUPKTRCV	Packets received. Not supported for type 6.
BUBYTESND	Data bytes sent. Not supported for type 6.
BUBYTERCV	Data bytes received. Not supported for type 6.
BUMAXRATE	Maximum byte rate. The estimated maximum rate that data can be both sent and received in bytes per second.
BUDATA1	The meaning of this field depends on the type (BUTYPE) field: •Type 4: Reserved. •Type 6: Command bytes sent.
BUDATA2	The meaning of this field depends on the type (BUTYPE) field: •Type 4: Reserved. •Type 6: Command bytes received.

Column name	Description
DSPTROP	The path total read operations reports the number of read requests that are received by internal machine functions, which is not the same as the device read operations reported in the DSDROP field.
DSPTWOP	The path total write operations reports the number of write requests that are received by internal machine functions, which is not the same as the device write operations that are reported in the DSDWOP field.
DSWWNN	The worldwide node name is a unique identifier that represents the external storage subsystem that the disk belongs to. This value is null for non-external disks.

QAPMDISKRB (microseconds)			QAPMDISK (milliseconds)
Bucket	>	<	Bucket	>	<
1	0	15	1	>0	1
2	15	250
3	250	1,000
4	1,000	4,000	2	2	16
5	4,000	8,000
6	8,000	16,000
7	16,000	64,000	3	16	64
8	64,000	256,000	4	64	256
9	256,000	500,000	5	256	1024
10	500,000	1,024,000
11	1,024,000		6	1024

Column name	Description
INTNUM	Interval number: The nth sample database interval based on the start time specified in the Create Performance Data (CRTPFRDTA) command.
DATETIME	Interval date (yymmdd) and time (hhmmss): The date and time of the sample interval.
INTSEC	Elapsed interval seconds: The number of seconds since the last sample interval.
ETMPORT	Port resource name.
ETMFTG	Frames transmitted without error.
ETMFRG	Frames received without error.
ETMCRE	CRC error: Checksum errors detected by the receiver.
ETMEXR	More than 16 retries: Frame unsuccessfully transmitted due to excessive retries.
ETMOWC	Out of window collisions: Collision occurred after slot time of channel elapsed.
JBINDCPU	The amount of unscaled processor time (in µs) that represents the work that is done solely within this thread without regard for how server task work is charged.
ETMALE	Alignment error: Inbound frame contained non-integer number of bytes and a CRC error.
ETMCRL	Carrier loss: Carrier input to the chipset on the IO adapters is false during transmission.
ETMDIF	Discarded inbound frames: Receiver discarded frame due to lack of AIF entries.
ETMROV	Receive overruns: Receiver has lost all or part of an incoming frame due to buffer shortage.
ETMMEE	Memory error: The chipset on the IO adapters is the bus master and did not receive ready signal within 25.6 microseconds of asserting the address on the DAL** lines.
ETMSQE	Signal quality error: Signal indicating the transmit is successfully complete did not arrive within 2 microseconds of successful transmission.
ETMM1R	More than one retry to transmit: Frame required more than one retry for successful transmission.
ETM1R	Exactly one retry to transmit: Frame required one retry for successful transmission.
ETMDCN	Deferred conditions: The chipset on the IO adapters deferred transmission due to busy channel
ETMBRV	Total MAC bytes received ok: This contains a count of bytes in frames that are successfully received. It includes bytes from received multicast and broadcast frames. This number includes everything from destination address up to but excluding FCS. Source address, destination address, length or type, and pad are included.
ETMBTR	Total MAC bytes transmitted ok: Total number of bytes transmitted successfully. This number includes everything from destination address up to but excluding FCS. Source address, destination address, length or type, and pad are included.
ETMTFD	Transmit frames discarded.
ETMUPF	Unsupported protocol frames: Number of frames that were discarded because they specified an unsupported protocol. This count is included in the discarded inbound frames counter.

Cache memory access: On Power Systems, all of the processor cores on any chip can access any of the cache memory in the entire system. The management of the relationship between the processor or “node” where a task runs and the “nodal” location where that task finds its data is called Memory Resource Affinity.

Column name	Description
JBNFHN	The identifier of a resource affinity domain that this software thread or task is associated with. A thread or task is associated with the resource affinity domain at a create time, but the operating system can decide to move it to another resource affinity domain later.
JBNFLVL	The resource affinity level specifies the relative strength of the binding between a thread and the internal machine resources with which it has affinity (processors and main storage). The strength is expressed as: •X’00’ = Processor normal, main storage normal •X’01’ = Processor normal, main storage high •X‘10’ = Processor high, main storage normal •X’11’ = Processor high, main storage high •X’03’ = Processor normal, main storage none •X’20’ = Processor low, main storage normal
JBNFGRP	The identifier of a resources affinity group or resource affinity domain. This identifier specifies how threads or tasks are related to other threads or tasks in their use of internal machine processing resources, or how they are related to specific resource affinity domains.
JBNFHNC	The amount of processor time that is used by the thread on the resource affinity domain that this thread is associated with. The time is reported in internal model-independent units. This time is called the Local Dispatch Time.
JBNFFNC	The amount of processor time that is used by the thread on resource affinity domains other than the one that this thread is associated with, but within the same group. The time is reported in internal model-independent units. This time is called the Non-Local Dispatch Time.
JBNFHNP	The number of 4-K page frames that are allocated for this thread during this interval from the resource affinity domain that this thread is associated with. These frames are called Local page frames.
JBNFFNP	The number of 4-K page frames that are allocated for this thread during this interval from resource affinity domains other than the one that this thread is associated with, but within the same group. These frames are called Non-local page frames.
JBTNAME	Identifies the name of secondary thread at sample time. The field is blank for primary threads, tasks, and unnamed secondary threads.
JBSLTCNT	If the short lifespan entry count is greater than zero, the entry does not represent a particular task or secondary thread. Instead, it is a special record that is used to report data that is accumulated for tasks and threads whose lifespan was shorter than the reporting threshold that was in effect when the collection started. Short lifespan tasks are reported for the processor node that they were associated with and short lifespan secondary threads are reported for the job to which they belong.
JBSACPU	The accumulated job scaled processor time that is charged (in microseconds). The accumulated scaled interval processor time that is charged for all threads of the job since the job started. This field is provided for primary threads only.
JBINDCPU	The amount of unscaled processor time (in µs) that represents the work that is done solely within this thread without regard for how server task work is charged.
JBSINDCPU	Thread scaled processor time that is used (in microseconds). The amount of scaled processor (in µs) time that represents the work that is done solely within this thread without regard for how server task work is charged.
JBCPUWC	The elapsed processor time (in µs) that a task runs.
JBVPDLY	The elapsed delay time (in microseconds) because virtualization for a task when it was running. The virtual processor delay time includes virtual processor thread wait event time, virtual processor thread wait ready time, and virtual processor thread dispatch latency.
JBSEIZECNT	The number of seizes held by this thread at the time that the data was sampled.
JBPSLCKCNT	The number of process scoped locks that are held by this thread at the time that the data was sampled.
JBTSLCKCNT	The number of thread scoped locks that are held by this thread at the time that the data was sampled.
JBTSRCDLCK	The number of thread scoped database record locks held by this thread at the time that the data was sampled.
JBNFOGDT	The amount of processor time that is used by the thread in a resource affinity group other than the one that this thread is associated with. The time is reported in internal model-dependent units.
JBNFOGMA	The number of 4-K page frames that are allocated for this thread during this interval from a resource affinity group other than the one that this thread is associated with.
JBFLDR2	Workload capping group delay time (in microseconds). The amount of time that this thread could not be dispatched because of workload capping.
JBFLDR3	Workload capping group. The identifier for the workload capping group that this thread belonged to at the time these data were sampled. A value of zero is reported when no group was assigned.
JBSQLCLK	SQL clock time. The amount of clock time (in microseconds) this thread has spent running work that was done on behalf of an SQL operation.
JBSQLCPU	Thread unscaled SQL CPU time used. The amount of unscaled processor time (in microseconds) this thread has used running work that was done on behalf of an SQL operation
JBSQLSCPU	Thread scaled SQL processor time used. The amount of scaled processor time (in microseconds) that this thread used running work that was done on behalf of an SQL operation.
JBSQLDBR	SQL synchronous database reads. The total number of physical synchronous database read operations that are done on behalf of an SQL operation.
JBSQLNDBR	SQL synchronous nondatabase reads. The total number of physical synchronous non-database read operations that are done on behalf of an SQL operation.
JBSQLDBW	SQL synchronous database writes. The total number of physical synchronous database write operations that are done on behalf of an SQL operation.
JBSQLNDBW	SQL synchronous nondatabase writes. The total number of physical synchronous non-database write operations that are done on behalf of an SQL operation.
JBSQLADBR	SQL asynchronous database reads. The total number of physical asynchronous database read operations that are done on behalf of an SQL operation.
JBSQLANDBR	SQL asynchronous nondatabase reads. The total number of physical asynchronous non-database read operations that are done on behalf of an SQL operation.
JBSQLADBW	SQL asynchronous database writes. The total number of physical asynchronous database write operations that are done on behalf of an SQL operation.
JBSQLANDBW	SQL asynchronous nondatabase writes. The total number of physical asynchronous non-database write operations that are done on behalf of an SQL operation.
JBHSQLSTMT	Number of high-level SQL statements. The number of high-level SQL statements that run during the Collection Services time interval. This count includes only initial invocation of the independent SQL statements. It does not include dependent SQL statements started from within another SQL statement. This count also includes initial invocation of independent SQL statements that failed to run successfully.

Column name	Description
SMPOOLID	Shared memory pool identifier. The identifier of the shared memory pool that this partition is using.
SMWEIGHT	Memory weight. Indicates the variable memory capacity weight that is assigned to the partition. Valid values are hex 0 - 255. The larger the value, the less likely this partition is to lose memory.
SMREALUSE	Physical real memory used. The amount of shared physical real memory, in bytes, that was being used by the partition memory at the sample time.
SMACCDLY	Real memory access delays. The number of partition processor waits that occurred because of page faults on logical real memory.
SMACCWAIT	Real memory access wait time. The amount of time, in milliseconds, that partition processors waited for real memory page faults to be satisfied.
SMENTIOC	Entitled memory capacity for I/O. The amount of memory, in bytes, assigned to the partition for usage by I/O requests.
SMMINIOC	Minimum entitled memory capacity for I/O. The minimum amount of entitled memory, in bytes, needed to function with the current I/O configuration.
SMOPTIOC	Optimal entitled memory capacity for I/O. The amount of entitled memory, in bytes, that allow the current I/O configuration to function without any I/O memory mapping delays.
SMIOCUSE	Current I/O memory capacity in use. The amount of I/O memory, in bytes, currently mapped by I/O requests.
SMIOCMAX	Maximum I/O memory capacity used. The maximum amount of I/O memory, in bytes, that was mapped by I/O requests since the partition last had an IPL or the value was reset by an explicit request.
SMIOMDLY	I/O memory mapping delays. The cumulative number of delays that occurred because insufficient entitled memory was available to map an I/O request since the partition last underwent an IPL.
MPACCDLY	Pool real memory access delays. The number of virtual partition memory page faults within the shared memory pool for all partitions.
MPACCWAIT	Pool real memory access wait time. The amount of time, in milliseconds, that all partitions processors spent waiting for page faults to be satisfied within the shared memory pool.
MPPHYMEM	Pool physical memory. The total amount of physical memory, in bytes, assigned to the shared memory pool.
MPLOGMEM	Pool logical memory. The summation, in bytes, of the logical real memory of all active partition active partitions that are served by the shared memory pool.
MPENTIOC	Pool entitled I/O memory. The summation, in bytes, of the I/O entitlement of all active partitions that are served by the shared memory pool.
MPIOCUSE	Pool entitled I/O memory in use. The summation, in bytes, of I/O memory that is mapped by I/O requests from all active partitions that are served by the shared memory pool.

Column name	Description
SYPTWAIT	Virtual processor thread wait event time. The elapsed time in microseconds that blocked threads of the partition's virtual processors were waiting for an event that caused them to become ready to run.
SYPTREADY	Virtual processor thread wait ready time. The elapsed time in microseconds that ready to run threads of the partition's virtual processors waited to be dispatched when entitled capacity was exhausted.
SYPTLATEN	Virtual processor thread dispatch latency. The elapsed time in microseconds that ready to run threads of the partition's virtual processors waited to be dispatched when entitled capacity was not exhausted and a physical processor was not available.
SYPTACT	Virtual processor thread active time. The elapsed time in milliseconds summed for all threads of a virtual processor for the time that the virtual processor is active in the partition. A virtual processor that is active is one that is varied on; a virtual processor that is not active is either varied off or not installed.
SYPTIDLE	Virtual processor thread idle time. The elapsed time in milliseconds summed for all threads of a virtual processor for the time that thread is idle in the partition. A processor thread that is idle is one that is varied on and running the partition's idle loop.
SYPTINTR	Virtual processor thread interrupt time. The elapsed time in milliseconds summed for all threads of a virtual processor for the time that thread is handling interrupts in the partition.
SYFRMCPU	Processor firmware time used (in microseconds). The amount of processor firmware time that is used by this partition.
SYFRMSCPU	Processor scaled firmware time that is used (in microseconds). The amount of scaled processor firmware time that is used by this partition.
SYFRMSCPU	Processor scaled firmware time that is used (in microseconds). The amount of scaled processor firmware time that is used by this partition.
SYPFOLDSW	Identifies the current state of the processor folding1 switch, where: •Blank means data not available •“0” off •“1” on •“2” system controlled
SYPFOLDST	The current state of processor folding, where: •Blank means data not available •“0” disabled •“1” enabled
SYEMMAJCDE	Energy management major code2.
SYEMMINCDE	Energy management minor codeb.
SYEMATTR	Energy management attributes. Bit 0 identifies the power draw limit type (0 = soft, 1 = hard).
SYEMPWRLMT	Energy management power draw limit in watts.

---

Column name	Description
INTNUM	Interval number. The nth sample database interval that is based on the start time that is specified in the Create Performance Data (CRTPFRDTA) command.
DATETIME	Interval date and time. The date and time of the sample interval.
INTSEC	Elapsed interval seconds. The number of seconds since the last sample interval.
SWGROUP	Group ID. The identifier for the workload group.
SWGNAME	Group Name. The name that is assigned to the workload group when it is allocated by License Management.
SWPRCASN	Processors assigned. The maximum number of processors that can be used concurrently by all threads of all processes that are associated with the workload group. This value is the value that is associated with the group at the time data was sampled.
SWPRCAVL	Processor time available (in microseconds). The amount of processor time that this group had available to it based on the number of processors that are assigned to the group over time.
SWPRCUSE	Processor unscaled time used (in microseconds). The amount of unscaled processor time that is used within the threads that are assigned to this group. This value does not include the time that charged to a thread by server tasks.
SWSPRCUSE	Processor scaled time that is used (in microseconds). The amount of scaled processor time that is used within threads that are assigned to this group. This value does not include the time that is charged to a thread by server tasks.
SWDELAY	Dispatch latency time. The amount of time ready to run threads could not be dispatched because of the group's maximum concurrent processor limit.
SWPRCADD	Processes added. The number of process instances that became associated with this group during the interval.
SWPRCRMV	Processes removed. The number of process instances that were disassociated from this group during the interval.

Column name	Description
TPRDS	Number of reads
TPWRTS	Number of writes
TPBRD	Bytes read
TPBWRT	Bytes written
TPWREQ	Time spent waiting for a request from the client (in milliseconds)
TPWRESP	Time spent waiting for a response from the drive (in milliseconds)
TPSFMCMD	Space by file mark commands
TPFLMRKSPC	File marks spaced
TPSBCMD	Space block commands
TPBLCKSPC	Blocks spaced
TPWFMCMD	Write file mark commands
TPFLMRKWRT	File marks written
TPSEODCMD	Space to EOD commands
TPWBCMD	Write buffer commands
TPRELEASES	Release commands
TPREWINDS	Tape rewinds
TPUNLOADS	Tape unloads
TPSTPOSCMD	Set tape position commands
TPRDPOSCMD	Read tape position commands

Column name	Description
INTNUM	Interval number. The nth sample database interval that is based on the start time in the Create Performance Data (CRTPFRDTA) command.
DATETIME	Interval date (mmddyy) and time (hhmmss). The date and time of the sample interval.
INTSEC	Elapsed interval seconds. The number of seconds since the last sample interval.
SITYPE	Record type. Always 3 for the values shown in this table.
SIDID	Internal record identifier. Always 1 for the values shown in this table.
SIDATA01	Total TLBIEs
SIDATA02	The number of ticks of the time base special purpose register spent processing TLBIEs this interval. On POWER7, there are 512 ticks in a microsecond. Divide the value in this record by 512 to get the total time spent processing TLBIEs in microseconds. Note: SIDATA02 / SIDATA01 = “Average TLBIE time”
SIDATA03	TLBIE stats start time. Not used by iDoctor.
SIDATA04	Total TLBIEs with a duration < 1 microsecond. Note: On fields SIDATA04-SIDATA09, divide by 1,000,000 to get the value in millions.
SIDATA05	Total TLBIEs with a duration between 1 and < 10 microseconds.
SIDATA06	Total TLBIEs with a duration between 10 and < 1000 microseconds.
SIDATA07	Total TLBIEs with a duration between 1 and < 10 milliseconds.
SIDATA08	Total TLBIEs with a duration between 10 and < 100 milliseconds.
SIDATA09	Total TLBIEs with a duration > 100 milliseconds.
SIDATA10	Average time that is spent processing TLBIEs (in ticks) in the last 10 milliseconds. Divide by 512 to get this value in microseconds.
SIDATA11	Average time that is spent processing TLBIEs (in ticks) in the last 100 milliseconds. Divide by 512 to get this value in microseconds.
SIDATA12	Average time that is spent processing TLBIEs (in ticks) in the last 1 second. Divide by 512 to get this value in microseconds.
SIDATA13	Average time that is spent processing TLBIEs (in ticks) in the last 10 seconds. Divide by 512 to get this value in microseconds.
SIDATA14	Average time that is spent processing TLBIEs (in ticks) in the last 100 seconds. Divide by 512 to get this value in microseconds.
SIDATA15	Average time that is spent processing TLBIEs (in ticks) in the last 1000 seconds. Divide by 512 to get this value in microseconds.

Column name	Description
INTNUM	Interval number. The nth sample database interval that is based on the start time in the Create Performance Data (CRTPFRDTA) command.
DATETIME	Interval date (mmddyy) and time (hhmmss). The date and time of the sample interval.
INTSEC	Elapsed interval seconds. The number of seconds since the last sample interval.
SITYPE	Record type. Always 3 for the values that are shown in this table.
SIDID	Internal record identifier. Always 2 for the values that are shown in this table.
SIDATA01	Total add HPTE requests.
SIDATA02	Total remove HPTE requests.
SIDATA03	Total set HPTE change bit requests.
SIDATA04	Total set HPTE storage key requests.
SIDATA05	Total set address compare bit requests.
SIDATA06	Total set HPTE page protect bit requests.
SIDATA07	Total get HPTE status bit requests.
SIDATA08	Total set HPTE reference bit requests.
SIDATA09	Current interval start time (not used by iDoctor.)
SIDATA10	Maximum paced spin wait time. Divide by 512 to get this value in microseconds.
SIDATA11	Total paced spin wait time. Divide by 512 to get this value in microseconds. Note: SIDATA11 / SIDATA12 = “Average paced spin wait time”
SIDATA12	Total paced spin waits TLBIEs.
SIDATA13	Maximum paced time. Divide by 512 to get this value in microseconds.
SIDATA14	Total paced time. Divide by 512 to get this value in microseconds.
SIDATA15	Total paced TLBIEs. Note: SIDATA14 / SIDATA15 = “Average paced TLBIE time”

Important: Major enhancements have been made to IBM Navigator for i and the Performance Data Investigator, so ensure that you have the latest group PTFs installed for the following groups: •HTTP Server group, PTF SF99368 •Java group, PTF SF99572 •Database group, PTF SF99701 •Performance Tools group, PTF SF99145