MonitorWD Syntax Elements - Teradata Vantage

PPId

This field is obsolete and returns a value of zero.

PGId

This field returns the pWDid value.

VprType

Type of vproc:

• AMP
• PE
• MISC

WDId

WD ID. On SLES 11 or later systems, TASM Workloads rule is enabled. On SUSE Linux Enterprise Server 11 or later systems, TASM Workloads rule is enabled.

AGId

This field is obsolete and returns a value of zero.

RelWgt

This field is obsolete and returns a value of zero.

NumProcs

Average number of tasks of online nodes.

The field is the result of:

NumProcs = SUM of (NumTasks-i) / N

where:

• NumTasks-i is the number of tasks assigned to the WD at the end of the reporting period.
• i varies from 1 to N, where N is the number of online nodes.

The NumProcs field is the NumTasks field in the PM/API MONITOR WD request.

QWaitTime

Total wait time in milliseconds that work requests waited on an input queue before being serviced.

QWaitTimeMax

Maximum time in milliseconds that work requests waited on an input queue before being serviced.

The field is the result of:

QWaitTimeMax = MAX (QWaitTimeMax-i)

where:

• QWaitTimeMax-i is QWaitTimeMax in each online node.
• i varies from 1 to N, where N is the number of online nodes.

CPUUserPct

Weighted average of CpuUserPct of each node.

The field is the result of:

CpuUserPct = Sum of (CpuUserPct-i* ScalingFactor-i) / Sum of (ScalingFactor-i)

where:

• CpuUserPct-i is calculated as:

  (CPUUServAwt + CPUUServDisp + CPUUServMisc + CPUUExecAwt + CPUUExecDisp + CPUUExecMisc) * 100 /(NCPUs*Centisecs*10)

  NCPUs is the number of CPUs in the node.

• i varies from 1 to N, where N is the number of online nodes.
• ScalingFactor-i is the node CPU normalization factor in each node.

The CPU times are in milliseconds.

The Parser CPU times are included in the Dispatcher CPU times.

WorkMsgMaxDelay

General indicator only. This field is result of the following calculation:

WorkMsgMaxDelay = MAX (WorkMsgMaxDelay-i)

where:

• WorkMsgMaxDelay-i is calculated in each online node as:

  WorkMsgsendDelayMax + WorkMsgReceiveDelayMax

• i varies from 1 to N, where N is the number of online nodes.

WorkMsgMaxDelay does not represent the subtotal of the same message on the send and receive side.

WorkTypeInuseMax

Total of the AMP Worker Task (AWT) columns:

WorkTypeInuseMax = MAX (WorkTypeInuseMax-i)

where:

• WorkTypeInuseMax-i is the sum of WorkTypeMax00 through WorkTypeMax15 in each node.
• i varies from 1 to N, where N is the number of online nodes.

WorkTimeInuseAvg

Average number of AWTs used. This field is result of:

WorkTimeInuseAvg = SUM of (WorkTimeInuse-i) / N

where:

• WorkTimeInuse-i is calculated in each online node as:

  WorkTimeInuse/(Centisecs * 10 * NCPUs)

  NCPUs is the number of CPUs in the node.

• i varies from 1 to N, where N is the number of online nodes.

This value is available in the ResSpsView view as AwtUsedAvg.

IODelay

Number of I/Os that are delayed. This field is result of:

ProcBlksFsgRead + ProcBlksFsgWrite + ProcBlksFsgNIOs

IODelayTime

Total time the I/O is delayed for. This field is the result of:

ProcWaitFsgRead + ProcWaitFsgWrite + ProcWaitFsgNIOs

PhysicalRead

Number of physical reads performed for this period. This field is the result of:

FilePDbAcqReads + FilePDbPreReads + FilePCiAcqReads + FileSDbAcqReads + FileSCiAcqReads + FileTJtAcqReads + FileAPtAcqReads + FilePCiPreReads + FileSDbPreReads + FileSCiPreReads + FileTJtPreReads + FileAPtPreReads

PhysicalReadKB

Number of physical reads in KB performed for this period. This field is result of:

FilePDbAcqReadKB + FilePDbPreReadKB + FilePCiAcqReadKB + FileSDbAcqReadKB + FileSCiAcqReadKB + FileTJtAcqReadKB + FileAPtAcqReadKB + FilePCiPreReadKB + FileSDbPreReadKB + FileSCiPreReadKB + FileTJtPreReadKB + FileAPtPreReadKB

PhysicalWrite

Number of physical writes performed for this period. This field is result of:

FilePDbFWrites + FilePCiFWrites + FileSDbFWrites + FileSCiFWrites + FileTJtFWrites + FileAPtFWrites

PhysicalWriteKB

Number of physical writers in KB performed for this period. This field is result of:

FilePDbFWriteKB + FilePCiFWriteKB + FileSDbFWriteKB + FileSCiFWriteKB + FileTJtFWriteKB + FileAPtFWriteKB

LogicalRead

Number of logical reads performed for this period. This field is result of:

FilePDbAcqs + FilePDbPres + FilePCiAcqs + FileSDbAcqs + FileSCiAcqs + FileTJtAcqs + FileAPtAcqs + FilePCiPres + FileSDbPres + FileSCiPres + FileTJtPres + FileAPtPres

LogicalReadKB

Number of logical reads in KB performed for this period. This field is result of:

FilePDbAcqKB + FilePDbPresKB + FilePCiAcqKB + FileSDbAcqKB + FileSCiAcqKB + FileTJtAcqKB + FileAPtAcqKB + FilePCiPresKB + FileSDbPresKB + FileSCiPresKB + FileTJtPresKB + FileAPtPresKB

LogicalWrite

Number of logical writes performed for this period. This field is result of:

FilePDbDyRRels + FilePCiDyRRels + FileSDbDyRRels + FileSCiDyRRels + FileTJtDyRRels + FileAPtDyRRels

LogicalWriteKB

Number of logical writes in KB performed for this period. This field is result of:

FilePDbDyRRelKB + FilePCiDyRRelKB + FileSDbDyRRelKB + FileSCiDyRRelKB + FileTJtDyRRelKB + FileAPtDyRRelKB

ExtraField1

ExtraField2

ExtraField3

ExtraField4

This field is not used.

VPId

Virtual partition ID.

WaitIO

Number of milliseconds tasks in WD waited for I/O over the reporting period.

WaitIO is updated when the wait for I/O is completed.

WaitOther

Number of milliseconds tasks in WD waited for reasons other than I/O over the reporting period (for example, a task waiting for a message).

WaitOther is updated when wait is completed.

CPURunDelay

Number of milliseconds tasks in the WD sat in the CPU runqueue waiting to run over the reporting period.

This data can be used in determining demand for the virtual partition and Workload Share Percent. The Workload Share Percent* is a workload management method. If the CPU and I/O percentages for a virtual partition or WD are below their relative share values and the CPURunDelay values are low, there was insufficient demand to meet the share percentage. If the CPURunDelay values are high, higher tier SQL requests were allocated more resources so that there were insufficient resources remaining to allocate to SQL requests in this WD to meet its relative share.

A virtual partition divides a system so that a percentage of resources are allocated to a collection of workloads. A virtual partition can consist of WDs from all management methods.

IOSubmitted

Number of I/Os submitted on behalf of this WD.

IOSubmittedKB

KB of I/O submitted on behalf of this WD.

IOCompleted

Number of AgeOut Now data blocks not to keep in memory (fsgcache) and to be written to disk.

IOCompletedKB

KB of AgeOut Now data blocks not to keep in memory (fsgcache) and to be written to disk.

IOCriticalSubmitted

Number of I/Os submitted with critical status. These I/Os run at top priority instead of being based on the I/O priority of the SQL request.

IOCriticalSubmittedKB

KB of I/O submitted with critical status. These I/Os run at top priority instead of being based on the I/O priority of the SQL request.

DecayLevel1IO

Number of times SQL requests in the WD reached decay level 1 due to I/O.

DecayLevel1IO is used for Timeshare WDs** only.

DecayLevel2IO

Number of times SQL requests in the WD reached decay level 2 due to I/O.

DecayLevel1IO is used for Timeshare WDs** only.

DecayLevel1CPU

Number of times SQL requests in the WD reached decay level 1 due to CPU.

DecayLevel1IO is used for Timeshare WDs** only.

DecayLevel2CPU

Number of times SQL requests in the WD reached decay level 2 due to CPU.

DecayLevel1IO is used for Timeshare WDs** only.

TacticalExceptionIO

Number of times SQL requests in the WD reached a tactical per-node exception due to I/O.

An exception, used only for Tactical WDs, is created for each Tactical WD***.

TacticalExceptionCPU

Number of times SQL requests in the WD reached a tactical per-node exception due to CPU.

TacticalExceptionCPU is used for Tactical WDs*** only.

* The Workload Share Percent Management Method workload is assigned a proportion of the resources that are available after allocations have been made for tactical workloads. The percentage of resources is divided equally between all requests running in the WD. For example, if the Workload Share Percent is 5% and there are five SQL requests, each SQL request gets 1% of the share resources.

** The Timeshare Workload Management Method workload can be assigned to one of four stepped access levels, Top, High, Medium, or Low. The higher access levels are given larger access rates than the lower levels. For example, an SQL request assigned to a Timeshare WD with a Top access level, which has an access rate of 8, gets eight times the amount of resources than an SQL request assigned to a Low access level.

Timeshare workloads are assigned resources remaining after all allocations have been made for tactical and Workload Share Percent workloads.

*** The Tactical Workload Management Method workload yields the fastest available response time and runs at the highest tier, preempting all resource needs of other tiers. This method is well suited for critical, short-running queries that require fast response times.