TPDC Manual (G06.24+)
Table Of Contents
- What’s New in This Manual
- About This Manual
- 1 TPDC Overview
- TPDC Components and Architecture
- Tandem Performance Data Collector (TPDC)
- Tandem Performance System File Label (TPSFL)
- Tandem Performance File Label (TPFL)
- Tandem Performance Load Measure (TPLM)
- Tandem Performance Load Data (TPLD)
- System Performance Analysis Using Measure (SPAM)
- Tandem Performance Deferred Processing (TPDP)
- Data Files (RWS, DWS, UWS)
- How TPDC Collects Data
- Example: Scheduling TPDC Data Collection
- TPDC Components and Architecture
- 2 Installing and Running TPDC
- 3 Using the Collected Data
- 4 Using the System Configuration Reports
- 5 Using the SPAM Performance Data Charts and Reports
- Overview of the Charts and Reports
- Performance Data Charts
- Acceleration Profiles of User Programs (AXLCHRT)
- Apportioned Program CPU Consumption (WKLOAD)
- Average CPU Utilization (ACUCHRT)
- Average Disc Utilization (ADUCHRT)
- Data Communications Line Overview (DATACOM)
- Disc Cache Performance (DCPCHRT)
- Disc File Block Splits Shares (BLKSPLT)
- Disc File I/O Activity Distribution (DFIODST)
- Disc File Overview (Alphabetic) (DFOATBL)
- Disc File Physical I/O Shares (DFPIOS)
- Disc File ‘Requests Blocked’ Shares (REQBLKD)
- Disc File System I/O Shares (DFSIOS)
- Disc I/Os per Second (DIOCHRT)
- Disc Response Time (DRTCHRT)
- EXPAND Communications Line Overview (EXPAND)
- Global Indicators (COVER)
- Logical Disc File I/O Wait Shares (LDFIOWT)
- SQL/MP Program Overview (SQLPROG)
- SQL/MP Statement Overview (SQLSTAT)
- Process Category Counts (PCOUNTS)
- Process/CPU Priority Distribution (PRIDIST)
- Program Disc File I/O Wait Shares (PGDIOWT)
- Program Logical File Open Shares (PGOPENS)
- Program Logical I/O Wait Shares (PGLIOWT)
- Program Overview (PGOVIEW)
- Program Spooler File I/O Wait Shares (PGSPLWT)
- Relative CPU Balance (RCBCHRT)
- Relative Disc Balance (RDBCHRT)
- TMF Overview (TMFVIEW)
- Performance Data Reports
- Busiest Disc Files (Logical I/O Wait Time, Overall) (IOWOVER)
- Busiest Disc Files (Logical I/Os, by Volume) (LIOBVOL)
- Busiest Disc Files (Physical I/Os, by Volume) (PIOBVOL)
- Busiest Disc Files (System I/Os, by Volume) (SIOBVOL)
- Busiest Processes (by CPU) (PBSYCPU)
- CPU Activity (CPUREPT)
- Disc Cache 1 (Activity, by Block Size) (CACHE1)
- Disc Process Activity (DPREPT)
- Disc Subsystem Activity (DSCREPT)
- EXPAND Network Line Activity (NETREPT)
- Global Indicators (COVER)
- Low Selectivity (Needle in a Haystack) SQL Table Access Report (HAYSTCK)
- Network System Activity (SYSREPT)
- New Process Creation Timeline (NEWPROC)
- PATHWAY System Processes (PATHWAY)
- Physical Disc I/O Activity (PDISCIO)
- Process Existence Timeline (PTIMLIN)
- Process Memory Utilization (by CPU) (PMEMCPU)
- Process Page Faults (by CPU) (PFLTCPU)
- Server Summary (from $RECEIVE File) (SERVICE)
- TMF Activity Report (TMFREPT)
- A Entity and Attribute Definitions
- B Cross-Reference of Enform Queries to Data Files
- C Enform Query Source File (QPGROUPS)
- Index

Using the SPAM Performance Data Charts and
Reports
Tandem Performance Data Collector (TPDC) Manual—526357-001
5-45
Relative CPU Balance (RCBCHRT)
Relative CPU Balance (RCBCHRT)
This 74-column chart displays the relative CPU balance. The measure of balance is
each CPU’s CPU-QTIME expressed as a percentage of the sum of all CPU-QTIME for
all CPUs (shown as Shr% in the chart). The chart displays the component parts of this
Shr%: memory queuing, other CPU queuing, and actual CPU usage.
This chart uses the GLOBINFO and CPU entities.
Parameters used are:
•
LID—specifies the Measure Load ID
•
SECTION—specifies the report section number
Relative CPU Balance Chart for System \CSO (Release D20)
--------------------------
The measure of balance is each CPU's CPU-QTIME, expressed as a percentage
of the sum of all CPUs' CPU-QTIMEs, and shown as 'Shr%' in this chart.
[ Exact Balance: Shr% = 25.0 per CPU ]
<--Shr% = %-age of all CPU Queueing & Usage--> Cum.
CPU TotQ MemQ Shr% 0 10 20 30 40 50 60 70 80 90 Shr%
--- ---- ---- ===== +----+----+----+----+----+----+----+----+----+ -----
1 .2 .0 37.4 |*****++++++++++++++ | 37.4
2 .1 .0 27.0 |*****+++++++++ | 64.4
0 .1 .0 21.8 |***++++++++ | 86.2
3 .1 .0 13.8 |*++++++ | 100.0
+----+----+----+----+----+----+----+----+----+
0 10 20 30 40 50 60 70 80 90
# = Memory Q'ing, * = CPU Q'ing, + = CPU Usage.
In a well-balanced, steady-state, system with adequate capacity for
instantaneous peaks, you would expect to see:
1. An almost-vertical right-hand edge to the chart,
2. NO pound signs [#], and
3. Fewer asterisks [*] than plus signs [+] on each line.
For faster CPU models (e.g. the NonStop (tm) Cyclone System), it may be
acceptable for the number of asterisks [*] to exceed the number of plus
signs [+], since each queued request can be serviced more rapidly.