TASM Columns | ResUsageSps Table | Teradata Vantage - 17.05 - TASM: Monitor WD Columns - Teradata Database

Teradata Vantage™ - Resource Usage Macros and Tables

prodname
Advanced SQL Engine
Teradata Database
vrm_release
17.00
17.05
created_date
June 2020
category
Administration
User Guide
featnum
B035-1099-170K
The Monitor WD columns provide RSS data to the following System PMPC APIs:
  • MONITOR WD request
  • MonitorWD function

For more information on these APIs, see Teradata Vantage™ - Application Programming Reference, B035-1090.

The Control AMP WorkTimeInuse and WorkTimeInuseMax columns described in the following table exclude uncompleted work time for the Control GDO AMP during recovery work. This will occur for rows with the RowIndex1 column equal to zero, but will typically not be noticed. Work time for recovery work that completes during the reporting period is reported for the Control AMP.

For more information on the RowIndex1 column, see Miscellaneous Housekeeping Columns.

Column Name Mode Description Data Type
ActiveSessions track Number of request-level workload management objects created. There is one object created per user request. For non-user work request, there is one object created per WD. FLOAT
CPURunDelay count Number of milliseconds tasks in the WD sat in the CPU run queue waiting to run over the reporting period.

Tasks may wait for access to a CPU regardless of the presence of COD or VP/WD Hard Limits. In addition, Tactical work is not immune to delayed access to CPU because the number of tasks running on a system often outnumbers the number of available CPUs. Delay can also be attributed to the running of tasks in higher priority System and Internal WDs, context switching, or excessive task activity in a WD.

This data can be used in determining demand for the virtual partition and WD Share 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.

It is more relevant and helpful to look at this value in relation to other WDs as opposed to looking at the value from one WD alone.

For more information on WD share workload management method, see Teradata® Viewpoint User Guide, B035-2206.

FLOAT
DecayLevel1IO count Number of times SQL requests in the WD hit decay level 1 due to I/O.

The values are reported under VprType MISC because the summed usage from PE and AMP go toward the usage that trips the decay. The values reported from either PE or AMP usage alone would not be accurate.

DecayLevel1IO is used for Timeshare WDs only. For more information on WD share workload management method, see Teradata® Viewpoint User Guide, B035-2206.
FLOAT
DecayLevel2IO count Number of times SQL requests in the WD hit decay level 2 due to I/O.

The values are reported under VprType MISC because the summed usage from PE and AMP go toward the usage that trips the decay. The values reported from either PE or AMP usage alone would not be accurate.

DecayLevel2IO is used for Timeshare WDs only. For more information on workload management methods, see Teradata® Viewpoint User Guide, B035-2206.
FLOAT
DecayLevel1CPU count Number of times SQL requests in the WD hit decay level 1 due to CPU.

The values are reported under VprType MISC because the summed usage from PE and AMP go toward the usage that trips the decay. The values reported from either PE or AMP usage alone would not be accurate.

DecayLevel1CPU is used for Timeshare WDs only. For more information on workload management methods, see Teradata® Viewpoint User Guide, B035-2206.
FLOAT
DecayLevel2CPU count Number of times SQL requests in the WD hit decay level 2 due to CPU.

The values are reported under VprType MISC because the summed usage from PE and AMP go toward the usage that trips the decay. The values reported from either PE or AMP usage alone would not be accurate.

DecayLevel2CPU is used for Timeshare WDs only. For more information on workload management methods, see Teradata® Viewpoint User Guide, B035-2206.
FLOAT
IOCompleted count Number of I/Os completed on behalf of this WD.
This column is valid on SLES 11 or later systems only.
FLOAT
IOCompletedKB count KB of I/O completed on behalf of this WD. FLOAT
IOCriticalSubmitted count Number of I/Os submitted with a MI Lock Priority. These I/Os go into a Teradata Scheduler queue and are not prioritized based on the type of management method. FLOAT
IOCriticalSubmittedKB count KB of I/O submitted with an MI Lock Priority. These I/Os go into a Teradata Scheduler queue and are not prioritized based on the type of management method. FLOAT
IOSubmitted count Number of I/Os submitted on behalf of this WD. FLOAT
IOSubmittedKB count KB of I/O submitted on behalf of this WD. FLOAT
NumRequests count Number of AMP Worker Task messages or requests that got assigned AMP Worker Tasks to them. FLOAT
NumTasks track Average number of tasks of online nodes. The column is the result of:

NumTasks = SUM of (NumTasks-i) / N

where:
  • NumTasks is the number of tasks assigned to the WD at the end of the reporting period.
  • i varies from 1 to N.
  • N is the number of online nodes.
FLOAT
QWaitTime count Total time in milliseconds that work requests waited on an input queue before being serviced. If the work requests are not delivered, they are not counted.

To calculate the average QWaitTime per request, divide by NumRequests, which is reported in DBC.ResSpsView as QWaitTimeRequestAvg.

FLOAT
QWaitTimeMax max Maximum time in milliseconds that work requests waited on an input queue before being serviced. FLOAT
ServiceTime count Time in milliseconds that work requests required for service.

To calculate an approximate average ServiceTime for each request during this period, divide the ServiceTime value by the NumRequests value, which is reported in DBC.ResSpsView as ServiceTimeRequestAvg.

The service time is the elapsed time from the time the message was received to the time the AMP Worker Task was released. This is the amount of time the AMP Worker Task was held through sleeps, CPU, I/O, and so on until it is released.

FLOAT
ServiceTimeMax max Maximum time in milliseconds that work requests required for service. FLOAT
TacticalExceptionIO count Number of times SQL requests in the WD hit a tactical per-node exception due to I/O.

The values are reported under VprType MISC because the summed usage from PE and AMP go toward the usage that trips the exception. The value reported from either PE or AMP usage alone would not be accurate.

An exception, used only for Tactical WDs, is created for each Tactical WD. For more information on workload management methods, see Teradata® Viewpoint User Guide, B035-2206.

FLOAT
TacticalExceptionCPU count Number of times SQL requests in the WD hit a tactical per-node exception due to CPU.

The values are reported under VprType MISC because the summed usage from PE and AMP go toward the usage that trips the exception. The value reported from either PE or AMP usage alone would not be accurate.

TacticalExceptionCPU is used for Tactical WDs only. For more information on workload management methods, see Teradata® Viewpoint User Guide, B035-2206.
FLOAT
WaitIO count 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.

FLOAT
WorkMsgReceiveDelay track Time for all messages not yet delivered at the end of each reporting period. This column is related to the QWaitTime column and represents a running total of delays attributed to the tasks that still have not been assigned an AMP Worker Task within this interval. When the task does receive an AMP Worker Task in a later interval, the time attributed here is counted again within QWaitTime of the interval where it was assigned an AMP Worker Task. FLOAT
WorkMsgReceiveDelayCnt track Number of messages that are still waiting for AMP Worker Tasks at the end of each reporting period. FLOAT
WorkMsgReceiveDelayMax track Maximum delay time in milliseconds for messages that are still in the work box at the end of each reporting period. FLOAT
WorkMsgSendDelay count Total time in milliseconds it takes to deliver work messages. FLOAT
WorkMsgSendDelayCnt count Number of messages that are delivered to the work box. FLOAT
WorkMsgSendDelayMax max Maximum time in milliseconds that it takes to deliver any single work message. FLOAT
WorkTimeInuse count Service time consumed by a WD during the current reporting period.
This is not the running sum of a WD that exists over multiple intervals.
FLOAT
WorkTimeInuseMax max Maximum service time of a single task in a WD that is running or has finished in the current reporting period. This includes time used during previous intervals for that task. FLOAT