Tandem Capacity Model (TCM) Manual HP Part Number: 522695-009 Published: August 2013 Edition: J06.03 and subsequent J-series RVUs, H06.03 and subsequent H-series RVUs, and G06.
© Copyright 2013 Hewlett-Packard Development Company, L.P. Confidential computer software. Valid license from HP required for possession, use or copying. Consistent with FAR 12.211 and 12.212, Commercial Computer Software, Computer Software Documentation, and Technical Data for Commercial Items are licensed to the U.S. Government under vendor’s standard commercial license. Warranty The information contained herein is subject to change without notice.
Contents About This Document...................................................................................11 New TCM Features and Microsoft Excel 2010............................................................................11 Enhancements...................................................................................................................11 Considerations..................................................................................................................
Process Category (PCAT) Statements.....................................................................................53 TITLE Statements.................................................................................................................58 File Defaults......................................................................................................................59 TXN Entries.................................................................................................................
Working With an Existing Performance Model...........................................................................88 Modifying the Planning Unit................................................................................................89 Modifying Percentile Response Times....................................................................................90 Using the Consumption Model.................................................................................................
Significant Rows..............................................................................................................123 Significant Columns.........................................................................................................124 History Charts......................................................................................................................124 Trend Chart...........................................................................................................
Screen Descriptions...............................................................................................................138 Capacity Baseline—Actual Screen..........................................................................................138 Significant Rows..............................................................................................................139 Significant Columns........................................................................................................
Narrow Column...............................................................................................................174 Print Tab..............................................................................................................................174 Print (F6).........................................................................................................................174 Page Setup....................................................................................................
Changing Expert Parameters..................................................................................................233 Changing Individual Priorities.................................................................................................234 Changing Passwords.............................................................................................................234 Checking an Apportionment Method.......................................................................................
Category Demand report.......................................................................................................253 Consumption model..............................................................................................................253 CPU conversion factors..........................................................................................................253 custom folder...........................................................................................................
About This Document Tandem Capacity Model (TCM) is a host-based and workstation-based capacity management tool that predicts the number of processors and logical disk volumes required for a well-tuned system, and estimates the expected response time and transaction throughput for OLTP and batch transactions. TCM works with MeasTCM. TCM uses data collected by Measure on HP NonStop host systems and summarized by MeasTCM.
◦ Choose the file to download, depending on the operating system installed (32 or 64 bit), and click Next. ◦ After the download is complete, run the downloaded file (.exe) to install the necessary files in the system. Supported Release Version Updates (RVUs) This manual supports J06.03 and all subsequent J-series RVUs, H06.03 and all subsequent H-series RVUs, and G06.14 and all subsequent G-series RVUs, until otherwise indicated in a replacement publication.
Changes to the 522695-007 Manual • Replaced most of the figures to show Excel 2007 screens and made the appropriate updates to the descriptions. • Changed prequisite manuals under “Related Information” (page 16). • Indicated, under “Functions and Features” (page 18), that both single-core and multi-core processors are supported. • Under “Understanding the Application” (page 21), changed UPDT to UPD.
• Updated the list of disk model numbers under DISKFACTOR Entry. • Added ACTUALDISK53 Entry as new example under DISKFACTOR Entry. Changes to the 522695-004 Manual The template control file has been updated and five new environmental entries have been described in Chapter 3, Using MeasTCM. Document Organization This document is organized as follows: Chapter 1: “Introduction” gives an overview of the product.
Notation Conventions General Syntax Notation This list summarizes the notation conventions for syntax presentation in this manual. UPPERCASE LETTERS Uppercase letters indicate keywords and reserved words. Type these items exactly as shown. Items not enclosed in brackets are required. For example: MAXATTACH Italic Letters Italic letters, regardless of font, indicate variable items that you supply. Items not enclosed in brackets are required.
{ } Braces A group of items enclosed in braces is a list from which you are required to choose one item. The items in the list can be arranged either vertically, with aligned braces on each side of the list, or horizontally, enclosed in a pair of braces and separated by vertical lines. For example: LISTOPENS PROCESS { $appl-mgr-name } { $process-name } ALLOWSU { ON | OFF } | Vertical Line A vertical line separates alternatives in a horizontal list that is enclosed in brackets or braces.
Publishing History Part Number Product Version Publication Date 522695-002 TCM G06 August 2002 522695-003 TCM G06 December 2002 522695-004 TCM G06 August 2004 522695-006 TCM H01 August 2008 522695-007 TCM H01 February 2012 522695-008 TCM H01 August 2012 522695-009 TCM H01 August 2013 HP Encourages Your Comments HP encourages your comments concerning this document. We are committed to providing documentation that meets your needs.
1 Introduction TCM is a workstation-based capacity planning tool for NonStop systems. Capacity planners can use TCM to estimate future system performance from current data. The TCM system is modeled in terms of transactions so that the results can be readily understood by a wide audience. TCM uses data from the Measure product, a performance monitoring product that runs on the HP NonStop system.
Assumptions To keep TCM simple and easy to use, TCM is based on the following assumptions: • Your HP NonStop system is well-tuned. (You can tune your HP NonStop systems by using the Guardian Performance Analyzer, HP product number T6942.) If the system is not well-tuned, response times might be higher than expected, more CPUs or disks might be necessary to support predicted workloads, or both. • TCM computes only the CPU and disk components of response time.
With the Performance model, you can explore possibilities by imposing constraints on the modeling process. You can impose constraints on either hardware configuration or performance factors. For instance, you can ask what will happen to response time if the average transaction rate doubles but the hardware configuration remains the same. Or you can ask what is the lowest-cost combination of CPUs and disk drives that will maintain an average response time below two seconds.
check workload apportionment methods that translate process-oriented performance data into transaction demands. In addition to the basic TCM functions, TCM administrators have access to the TCM administrative functions and can directly modify more data in the models than capacity planners. • Super users Super users can do everything that TCM administrators can do, and they also have the authority to resecure the model to gain access to all fields within TCM.
Figure 2 (page 22) shows a system-level process flow diagram for a sample application. Each oval in the figure represents a process category. This application has four transactions: standard (ST1), inquiry (INQ), big (BIG), and update (UPD). The application is described in the following list: • The server classes for each of the transaction types are ST1, INQ, BIG, and UPD. • The terminal control process (TCP) for ST1 is ST1_TCP. For INQ, BIG, and UPD, it is ST2_TCP.
Use MeasTCM to aggregate and analyze the samples into process category data in the form of comma separated values (CSV) files. When you are finished collecting performance data, download the CSV files created by MeasTCM to the custom folder on your workstation. For information about MeasTCM, see Chapter 3: “Using MeasTCM”. Apportioning Performance Data The fourth step in the capacity planning process is apportioning performance data.
Use TCM to: • Load the baseline into the Performance model • Modify scenarios and make projections • Create charts of possible scenarios Reporting Results The final step in the capacity planning process is reporting the results. Once you have determined the system resource needs, report the results to management. TCM helps you create a capacity plan using your Performance model projections and presentation-quality charts that are easy to read and understand.
2 TCM System Setup This chapter describes the system requirements for TCM and MeasTCM, provides installation procedures, explains how to find version information, discusses how TCM converts models from earlier TCM versions, and describes how to load models from other directories. System Requirements Before you begin using MeasTCM and TCM, make sure that you have the system requirements shown in Table 1 (page 25).
Figure 3 Installing TCM Flowchart Installing MeasTCM on the Host Install MeasTCM software files from the SUT. The MeasTCM installation subvolume uses about 175 KB of disk space. When you restore MeasTCM files from the SUT, the following files are located in the ISV: File name Description CNTLFILE Template control file MEASTCM MeasTCM program T6042ABD MeasTCM softdoc Installing TCM on the PC The instructions in this manual assume that your folder is named TCM.
2. 3. 4. requirements, see the latest TCM softdoc. If necessary, follow the procedures described in the application documentation to install the correct version of Windows, and Excel, or both. To determine whether you need to download any other files, see the README file on the ISV. Download any necessary files. If you have an earlier version of TCM installed on your PC, delete the TCM application files or put them in a different folder. Only one folder for TCM is allowed on a workstation.
For example, the original file bank.HST is renamed _ank.HST, and the converted file is now named bank.HST. A Microsoft Excel dialog box shows when the conversion is done. • The Performance Models (the .SPM and .WPM files) created by pre-G06 TCM product versions will not open (the Work with an Existing Model of the TCM Initial menu) in G06 TCM. To migrate these old Performance Models to this RVU: 1. Select TCM Initial Create a new Performance Model. 2.
3 Using MeasTCM MeasTCM provides the interface between Measure on the NonStop host system and TCM on your workstation. Running under TACL on the NonStop host system, MeasTCM summarizes and categorizes the performance data collected by Measure and formats it for use by TCM. MeasTCM reduces Measure performance data to a manageable form for capacity planning. Typically, MeasTCM reduces a Measure data file to just a few kilobytes—a size that contains all the performance data required by TCM.
Figure 4 MeasTCM Architecture MeasTCM Procedures To use MeasTCM: 1. Collect the Measure samples. 2. Draw a process flow diagram for your application. 3. Customize the MeasTCM control file. 4. Run MeasTCM to create the CSV file or files. Collecting Measure Samples For MeasTCM to extract and summarize Measure data, the Measure sample data must meet these requirements: • Data must be collected using a currently supported version of Measure.
• If you are modeling a batch job that runs when the online transaction application is running at a lower throughput or is shut down, create two models: ◦ One model for the online transaction application during its peak period ◦ One model for the batch job • If you are modeling a batch job that runs at the same time as the peak period for the online transaction application, the Measure sample should include all of the batch run and exclude the online transaction application startup and shutdown.
Figure 5 From Process Categories to Transactions You download the MeasTCM results into the WA model. During the workload apportionment processing, TCM divides categories into units that are allocated to transactions. After the workload apportionment process is complete, each transaction has a value for CPU seconds and disk seconds. If you obtain consistent CPU and disk seconds from the WA model for several Measure samples with a variety of transaction mixes, you have a reasonable WA model.
Figure 6 System-Level Process Flow Diagram of a Sample Application Processes in a category must not only recognize the same set of transaction types but also expend roughly the same proportion of effort on the transaction types they recognize. For example, suppose a requester and a server class processes T-1 and T- 2 (and no other types), and the requester expends twice as many CPU seconds on T-1 as T-2, but the ratio in the server class is 3:1. These process groups must be in separate categories.
Figure 7 Template Control File ! CNTLFILE T6042H01^ABD (24JUL13) ! This is a sample MeasTCM control file which defines MeasTCM ! processing requirements. This file must be modified to ! include actual data before it can be used.
! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! NSE-D,NSE-B,NSE-Q,NSE-T,NSE-K,NSE-I,NSE-S,NSE-M, NSE-W,NSE-AD,NSE-AB(2-CORE),NSE-AB(4-CORE)/NSE-AB, NSE-AE. [r] the OS release level,G H or J. This is optional. The default is G. xxxx.yy the new conversion factor for CPU type t; this is a positive number between 0001.00 and 9999.99. s [i] Type,Subtype of the CPU.
! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! 36 DISKFACTOR m r c s readseek [write seek] where: m the disk model number for which the default RPM, controller type of the disk will be overridden.
! ACTUALDISK44 m ! ! where: m the disk model number of subtype 44; ! Allowed values are: 4636, 4637, 4638. ! Note: Measure may not display the disk model number ! correctly. Physical checking of disk model number ! used during measurement is recommended. ! ! ACTUALDISK50 m [w] ! ! where: m the disk model number of subtype 50; ! Allowed value is: EO0200FBRVV ! ! [w] this is an optional field for WCE configuration; ! Allowed values are ON or OFF; By default it will be OFF.
! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! and 97, which identifies one of a set of counters used to summarize interval detail information B node name C D CPU number PIN number any or full Node name with "\") any ) any ) note that elements B,C,D are separated by a comma $E process name ( "$???????" or "*" means any) $F G H program volume name program subvol name program file name ( "$???????" or "*" means any) ( "????????" or "*" means any) ( "????????" or "*" means any) note that name is in
PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* * * * * * * * * * * * * * ! SQL/MX Processes ! PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * $system.system.sqlci2 $system.system.sqlcat $system.system.sqlcomp $system.system.audserv $system.system.sqlesp $system.system.
Section Description TITLE entries Provides a name and a type for each process category listed in the PCAT entries TXN (transaction) entries Contains a list of transaction names and types The first section, environmental entries, is required. For applications with OLTP transactions only, the three other sections are optional. For applications with batch transactions, the last section, TXN entries, is required.
application appears in the Workload Apportionment and Performance models on your workstation. CONVERTTO Entry The CONVERTTO entry represents the output CPU type. This entry is required when the measurement data file or files contains mixed CPU types. The syntax is: CONVERTTO t u Where: t is the output CPU type. u is the output NonStop Kernel product version. This entry is required when t (first parameter of CONVERTTO) is different from the CPU types contained in the measurement data files.
Table 2 CONVERTTO Values — Type, Model and Version (continued) Processor Type Processor Model NonStop Product Version S70K NSR-G G06, G05, G03, G02 S07K NSR-W G05, G03, G02 * Indicates multi-core processor When you use the CONVERTTO t u entry to normalize a measurement data file with mixed CPU types to an output CPU type, you can get a fractional number of CPUs. CPUFACTOR Entry CPUFACTOR is an optional entry that enables you to override the default CPU conversion factors.
Table 3 CPUFACTOR Values (continued) Processor Model Type, Subtype Name NonStop Product Version Default Values IPU Count NSR-Y 9,4 S86K G06 121.65 N.A. NSR-Z 9,1 S88K G06 152.83 N.A. NSE-Q 10,91 NS1200 H06 271.45 N.A. NSE-T 10,83 NS16200, NS16200CG H06 418.64 N.A. NSE-A 10,1 NS16000, NS16000CG H06 358.14 N.A. NSE-D 10,2 NS14000 H06 276.83 N.A. NSE-K 10,63 NS3000AC H06 231.06 N.A. NSE-I 10,51 NS5000T, NS5000CG H06 319.93 N.A.
NUMCSVS Entry The NUMCSVS entry specifies the number of CSV files you want MeasTCM to produce. If you do not specify a value, the default is 1. The syntax follows: NUMCSVS * | ALL | n where: * | ALL instructs MeasTCM to produce as many CSVs as possible. n is a positive integer. The number of CSV files generated depends on the start time, stop time, and intervals used in the input Measure data file or files in addition to the user-supplied CSVLO and CSVHI keyword parameters.
If the system has XP disk, then the changes in the control file for DISKCONVERTTO statement is necessary. The DISKCONVERTTO keyword requires a SAS disk model to be converted (and it does NOT accept JBOD disk and XP disks). Table 4 (page 45) lists the disks that can be used in DISKCONVERTTO statement.
This entry can contain only one disk model number. [w] specifies the write cache enable (WCE) option. This is an optional entry. If not specified, default values are used. Values can be ON or OFF, with the following defaults: • For disks that support WCE, the default value is OFF. • For disks that do not support WCE, this field is left blank.
Effective with SPR T6042H01^ABD and later, SAS disks are supported, with CLIM as the controller. However, XP disks are partially supported. If XP disks (subtype 52) are installed, then the DISKFACTOR directive must specify a SAS disk model name and subtype 51. However, the controller, read seek, and write seek times can reflect the actual values of the real disk types in the system.
Table 5 DISKFACTOR Model, Speed, Controller, Subtype and Seek Times (continued) 48 Disk Model Number Rotational Speed Controller Subtype Read Seek Time (ms) Write Seek Time (ms) 4510 5400 Ext 31 3.8 3.8 4560 5400 Ext 38 3.8 3.8 4570 7200 Ext 39 3.8 3.8 4580 7200 Ext 20 3.8 3.8 4590 7200 Ext 48 3.8 3.8 4604 7200 Snet 41 3.8 3.8 4608 7200 Snet 42 3.8 3.8 4609 10000 Snet 42 3.8 3.8 4618 7200 Snet 43 3.8 3.8 4619 15000 Snet 43 3.8 3.
Table 5 DISKFACTOR Model, Speed, Controller, Subtype and Seek Times (continued) Disk Model Number Rotational Speed Controller Subtype Read Seek Time (ms) Write Seek Time (ms) BF146DA47A 15000 FCSA 53 3.8 3.8 BF300DA47B 15000 FCSA 53 3.8 3.8 BF300D6188 15000 FCSA 53 3.8 3.8 BF450D6189 15000 FCSA 53 3.8 3.8 BF300DASTH 15000 FCSA 53 3.8 3.8 BF450DASTK 15000 FCSA 53 3.8 3.8 BF600DASTL 15000 FCSA 53 3.8 3.8 DG072BB975 10000 CLIM 51 4.0 4.
Table 5 DISKFACTOR Model, Speed, Controller, Subtype and Seek Times (continued) Disk Model Number Rotational Speed Controller Subtype Read Seek Time (ms) Write Seek Time (ms) DH072BAAKN 15000 CLIM 51 2.58 3.01 DG072BABCE 10000 CLIM 51 4.0 4.4 DG146BABCF 10000 CLIM 51 4.0 4.4 DG0146BARTP 10000 CLIM 51 4.0 4.4 DG0300BARTQ 10000 CLIM 51 4.0 4.4 EG0146FARTR 10000 CLIM 51 4.0 4.4 EG0300FARTT 10000 CLIM 51 4.0 4.4 EH0072FARUA 15000 CLIM 51 2.58 3.
ACTUALDISK42 Entry Two disk model numbers (4608 and 4609) can be associated with subtype 42. Therefore, ACTUALDISK42 is used to specify which of the two is in use. The syntax follows: ACTUALDISK42 m where: m is the actual disk model number (4608 or 4609) associated with the subtype 42. NOTE: Measure may not display the disk model number correctly. Physical checking of disk model number used during measurement is recommended.
NOTE: SSD disks does not have any moving parts. Therefore, rotational speed, read seek time, and write seek time are not applicable. Table 5 (page 47) lists the disks supported by TCM along with their model numbers, rotational speeds, controllers, and subtypes. NOTE: Measure may not display the disk model number correctly. Physical checking of disk model number used during measurement is recommended. ACTUALDISK51 Entry Effective with SPR T6042H01^ABD and later, SAS disks are associated with subtype 51.
ACTUALDISK53 Entry The ACTUALDISK53 entry is associated with disks of subtype 53 and is used to specify which disk is in use. The syntax follows: ACTUALDISK53 m where: m is the actual disk model number associated with the subtype 53. NOTE: Measure may not display the disk model number correctly. Physical checking of disk model number used during measurement is recommended.
I is the userid group (* means any). J is the userid user (* means any). In general, you should define these categories: • One category that includes all disk processes (except audit disk processes). Disk processes are assigned to one category only. • If the system is TMF-audited, one TMF category that includes the audit disk processes. (The default file assumes an audit disk name of $AUDIT.) • One category for each major server class.
Figure 8 PCAT Template File ! TMF processes PCAT01 *,*,4 * PCAT01 *,*,* $audit PCAT01 *,*,* $tmp PCAT01 *,*,* * PCAT01 *,*,* * $system.sys??.osimage $audi $system.sys??.* $tmp $system.sys??.* $system.sys??.tmf* $system.sys??.backout *,* *,* *,* *,* *,* ! TMF monitor ! audit disk ! $TMP ! >=D30 TMF ! backout process >D30 ! Requester (PATHWAY) processes PCAT02 *,*,* * $system.system.pathtcp2 PCAT02 *,*,* * $system.system.pathmon PCAT02 *,*,3 $zl?? $system.sys??.osimage PCAT02 *,*,* $zl?? $system.sys??.
PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 PCAT08 *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* *,*,* * * * * * * * * * * * * * * ! SQLMX Processes PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * PCAT09 *,*,* * $system.system.sqlci $system.system.sqlci2 $system.system.sqlcat $system.system.sqlcomp $system.system.
A process is potentially selected by a category if its attributes (PIN, program file name, and so on) match all the fields in at least one of the category’s PCAT statements. A process is only “potentially selected” because it might match PCAT statements in several categories. A process is selected by a PCAT statement if it matches all specifiers in the statement (an AND relationship).
Unspecified Data To ensure that no data is omitted, MeasTCM automatically summarizes all data that does not satisfy any other criteria in the file into category 98 with the default name of _OTHER and type O. All Discopen data that cannot be linked to process data goes into category 99, with the default name of _ERROR and type O. Figure 10 Multiple Category Matches TITLE Statements TITLE statements name a process category.
Table 6 TITLE Entry with Category Type and Description Title Entry Process Category Type Process Category Description TITLE01 T TMF (including audit disk processes) TITLE02 R Requester TITLE03 F Measure TITLE04 D Disk (excluding audit disk processes) TITLE05 O System TITLE06 S Server TITLE07 O OSS TITLE08 R SQLMP TITLE09 R SQLMX Effective with SPR T6042H01^ABD and later, OSS, SQLMP and SQLMX process categories are added to template control file.
The one-character transaction type code can be one of the following: • X for an OLTP transaction • B for a batch transaction • Q for a query transaction NOTE: Do not include a space between the one-character transaction type code and the transaction name. Including a space between them will cause an error. The transaction name can be up to 20 characters. If you do not specify TXN entries, TCM uses the name of the Server (S type) process categories as the default, and the TXN type is assumed to be X.
MeasTCM Syntax MeasTCM has the following syntax: [ RUN ] MEASTCM / [ run-option [ , run-option ] ] / cntlfile RUN is the TACL RUN command. Do not specify RUN if MeasTCM is installed in the $SYSTEM.SYSTEM subvolume. run-option is one of the following. This list describes commonly used RUN options. For a complete list of RUN options, see the TACL Reference Manual. CPU cpu-number specifies the number of the CPU in which MeasTCM is to run. cpu-number is an integer in the range of 0 through 15.
Figure 11 Run Summary Report MeasTCM Run Summary 24-JUN-1993 16:34:10 Measure Data File Summary: -------------------------1 Measure file name: $DATA17.TCMTEST1.MDATA1 System: \LILLI From time: 1991/09/17;10:00:00 To Time: 1991/09/17;12:00:00 Int.(Secs): 1200 2 Measure file name: $DATA03.TCMTEST1.MDATA2 System: \BELGIUM From time: 1991/09/17;10:00:00 To Time: 1991/09/17;12:00:06 Int.
Figure 12 Process Category Audit Report Process category audit report, application MDATA1 Measurement window: from 2002/09/17;10:20:00 to 2002/09/17;10:40:00 CSV Data file: CC20C001 Pcat System Program file name Process CP PIN Busy secs Res secs $VIRTUAL \BELGIUM $SYSTEM SYS00 OSIMAGE $VIRTUAL 0 1 .559 1200.00 1 1 .627 1199.95 2 1 .664 1199.92 3 1 .585 1199.94 \LILLI $SYSTEM SYS01 OSIMAGE $VIRTUAL 0 1 .000 1199.99 1 1 .000 1199.99 2.437 DISC \BELGIUM $SYSTEM SYS00 OSIMAGE $ADMIN 2 6 27.579 1199.89 33 1.
4 Getting Started With TCM This chapter provides some guidelines to help you start working with TCM. It also describes the TCM Initial menu, explains how TCM handles security issues, how to access TCM online help, and how to load TCM models created by other users. Guidelines Before you start using TCM, the following guidelines can help you use the application effectively. TCM Users TCM has three levels of users: • The capacity planner has access to basic commands and Performance model screens.
TCM distribution folder, but this architecture is recommended. Each folder should include all the files associated with the application: • Performance data files downloaded from the NonStop host system.
Figure 13 TCM Folder Structure TCM Initial Menu The TCM Initial menu gives you several choices. Descriptions of these choices follow.
Workload Apportionment Options Use the Workload Apportionment (WA) options to apportion sample data from processoriented to transaction-oriented information before doing what-if analysis by using the Performance model. Demonstrate Option When you choose the Demonstrate option, TCM displays an apportioned WA model that you can use for practice. You can make changes in the sample model, but you cannot save the changes.
1. From the “TCM Initial Menu” (page 66), select Work with an Existing WA Model or Work with an Existing Performance Model and then click OK. The Open PM Model or Open WA Model dialog box appears. 2. 3. 4. If you want to view all possible Workload Apportionment models, type *.SWA in the Model to Open field. Type *.SPM to see all possible Performance models. Select a file from the list or enter the file name you want to open, including its .SWA or .SPM extension. Click OK.
If you entered an incorrect, blank, or null password, your authorization level changes to capacity planner. TCM Administrator The TCM administrator is the second-highest level. It is password-protected. TCM administrators can do everything that capacity planners can do, and can also perform TCM administrative functions, such as creating apportionment tables. The default password is “admin.” This password is not case-sensitive, and both the WA and Performance models have the same default passwords.
superuser PM password to “orange.” Only the TCM administrator and superuser authorization levels can change the default passwords. CAUTION: If multiple users will be working with TCM, the default passwords should be changed to maintain system security. Changing Passwords To change passwords: 1. Log on as an administrator or super user. 2. Open a WA model or a Performance model. 3. Select Options > Change Password. The Change TCM Password dialog box appears. 4.
5 Creating a WA Model This chapter describes how to create a customized Workload Apportionment (WA) model for a new application (or set of applications). It is assumed in the following instructions that a Measure sample has been collected on the NonStop host system, and that a performance sample has been generated by MeasTCM, downloaded to the PC, and loaded into TCM. This chapter also includes a sample WA model for a non- Pathway application.
numbers in a given row. The numbers can be decimal fractions, such as 0.5. The numbers in one row have no effect on any of the other rows. The best way to read the table is across. In the following table, the CPU and disk consumption cost of the linehandler category is spread equally (on a per-transaction basis) between Withdrawal and Deposit. Deposit Server is charged entirely to Deposit. TCP is charged to Withdrawal and Deposit with relative weights of 2 and 1, respectively.
The Rcv/Send counts for each transaction name are multiplied by the user-entered coefficients and summed to produce a transaction count. In the example, the number of Deposit transactions is set equal to 1 times the messages received by the Deposit category. The following guidelines apply to the Transaction Count table: • Entries can be positive or negative. • Entries can be decimal fractions, such as 0.5.
NOTE: TCM enables you to modify specific cells in the WA workbook. TCM returns an error "To modify a protected cell, first remove protection using the Unprotect Sheet command (Review tab, Changes group)" when you try to edit any of the values in protected cells of the WA model workbook. Only TCM administrators are able to unprotect and edit admin cells. To edit the protected cells: 1. Change the Authorization level to "super" using the Options->Change Authorization Level button. 2.
WA Model Flowchart Figure 15 (page 75) outlines the major steps involved in creating a WA model. This section describes each of these steps. Figure 15 Creating a WA Model Loading Performance Data Into TCM TCM uses performance data samples and models as a basis for its capacity planning. Before you can model systems with TCM, you must download sample performance data to your PC and load the data into TCM.
same folder, and all CSV files must have the extension .CSV. TCM detects any truncated data and issues an error message. NOTE: All CSV files must have the extension .CSV. Loading Samples Into TCM To load a sample into TCM: 1. From the “TCM Initial Menu” (page 66), select Create a New WA Model. The Specify Custom Directory dialog box appears. 2. Specify the custom folder location of your CSV files. Select the custom folder of your CSV files and double-click the custom folder name.
Figure 16 Workload Apportionment Weights Table The default Apportionment Weights tables created by TCM provide an excellent starting point for fine tuning the WA model to your application’s particular architecture. However, to customize the WA model to best reflect an application’s characteristics, the TCM administrator might need to modify these tables.
NOTE: You can use a Transaction Count file instead of a Transaction Count table to count transactions processed by your application. For information on the Transaction Count file, see the subsection “Transaction Count Table Versus Transaction Count File” (page 120). If you are using a Transaction Count file, each entry in the Transaction Count table can be zero (0). 1. Open a WA model. 2. Log on as an administrator or super user.
1. Save the modified apportionment tables by selecting Save or Save As under the Find tab. The Save As option allows you to change the name of the apportionment tables. 2. 3. 4. 5. If you chose Save, the TCM Save dialog box appears. Choose the menu item you want and click Continue. • Select Workload > Apportionment Model files to save everything that is open in this session. • Select Apportionment Tables to save the apportionment tables.
1. 2. Check that the CPU and disk apportionments per transaction are in line with what you would expect. For example, you need to validate your apportionment method if your results show 2 seconds of CPU time for a transaction that you expect uses 500 milliseconds of CPU time. From the Find menu, select CPU, Disk Read, or Disk Write, under the Work Apportionment option, to view each apportionment in detail. Figure 18 (page 80) shows a sample CPU Workload Apportionment screen.
Figure 20 Capacity History Screen Saving a WA Model Save a model as a template for apportioning additional performance samples by saving the model’s apportionment tables and formulas. Create a new model if the apportionment method has been changed for an existing model or customized for one or more applications for the first time. To save a Workload Apportionment model with its existing name: 1. Select File > Save. 2. The TCM Save dialog box appears. Select the option you want and click Continue. 3. 4.
Checking an Apportionment Method Once an apportionment method has been created by modifying either the default apportionment tables of a new model or the apportionment tables of an existing model, and the apportionment formulas have been generated, you need to analyze the results to determine the reasonableness of the method.
3. 4. 5. 6. 7. 8. From the Expert tab to the TCM Expert Parameters, enter in Check History Over Last the number of months required to cover all dates when the samples were taken. Select Expert Update History with Selected CSVs. The Update History with Selected CSV Files dialog box appears. By default, the CSV files that have not yet been used to update capacity history appear in the Selected CSV Files list. Select a History database from the History Database list.
Non-ATM transaction is a transaction in which a bank customer performs an ATM function on an ATM that belongs to another bank. The Foreign transaction is a transaction in which another bank’s customer performs an ATM function on an ATM. Transactions enter the system at either the Device Handler or the Gateway Interface. A Router sends all transactions to Authorization. From Authorization, they return to the Router and are sent to Host Interface or Gateway Interface.
In each category, the per-transaction demand is estimated to be the same for all transaction types.
6 Creating a Performance Model This chapter describes how to create a Performance model (PM) for your application or set of applications. The Performance model enables you to ask questions about system performance under changing circumstances, such as transaction rates, response time constraints, and CPU type. As a capacity planner, you often have a number of options regarding the system’s response to increased workload. For example, you can add hardware to reduce CPU utilization and improve response time.
5. When you are finished, click OK. TCM performs the necessary calculations and then displays the Capacity Baseline—Actual screen. You are now ready to model scenarios. Figure 23 (page 88) shows a sample Capacity Baseline—Actual screen. If TCM encounters an unknown CPU in the WA model or CSV source file during calculation, it recalculates the sheet and saves the new file. To ensure proper Performance Model recalculation, however, you must enter the new CPU type into the CPU Conversion Factors screen.
Figure 23 Capacity Baseline—Actual Screen Working With an Existing Performance Model 1. 2. 3. Start TCM. The “TCM Initial Menu” (page 66) appears. Select Work with an Existing Performance Model and click OK. The Open PM Model dialog box appears so you can specify which Performance model to open and optionally which WA model to load into that Performance model. Select: • The custom folder.
WARNING! In the Excel 2003 version of TCM, the Custom Directory name text may appear corrupt. 4. When you finish, click Open. TCM shows the Capacity Baseline—Actual screen. You are now ready to model scenarios. Figure 23 (page 88) shows a sample Capacity Baseline—Actual screen. Modifying the Planning Unit Often a capacity planner needs to model transaction throughputs in units other than transactions per second (TPS).
Modifying Percentile Response Times TCM can provide the following results for OLTP transactions: average, 90th, 95th, or 99th percentile. The default is average. If you change from the default (average) to percentile response times, TCM uses percentile response time for OLTP transactions in all screens of the Performance model. For example, if the response time is 1 second and average response time is selected, the average response time of transactions is 1 second.
Actual data based on the workload-apportioned Measure data appears in the Actual column. Default constraints appear in the Constraint column. The configuration recommended by TCM appears in the Plan column. 3. 4. Enter constraints in the input fields in the Constraint column and other highlighted cells. Then press Return. After entering all of the changes you want into the highlighted input fields, press Return. Then select Options > Calculate Now.
Figure 25 Planning Timeline Model Screen 3. After entering all of the changes into the highlighted input fields, press Return. Then select Options Calculate Now. To finish entering your data, you must press Return before selecting Calculate Now. TCM calculates the values to reflect the changes. Throughput Trending You can use the Planning Timeline model to have TCM calculate throughput growth rate (or transaction per task growth rate) based on the existing trend in the history database.
3. 4. 5. 6. Select the transaction whose trend you want to measure from the Transaction list. If you want, modify the range of dates in the history database by entering new dates in the From and To fields. The default is the entire history database. Those dates are displayed in the fields. Click OK. Select Options > Calculate Now. Use the transaction trend in the historical database to update the Planning Timeline screen.
Figure 26 Planning Timeline Model Chart for Average Online Transaction Figure 27 Planning Timeline Model Chart for Batch Transaction 94 Creating a Performance Model
Printing Charts TCM creates its charts in black and white. TCM automatically produces the chart to be printed on a black and white printer. To print a chart: 1. Create a chart of the Planning Timeline model. 2. Update the Performance model data by selecting Options > Calculate Now. 3. Once the chart appears on the screen, select Print option under the Print tab. Saving the Model To save the Performance model with its existing name: 1. Select File > Save As. 2.
7 Modeling Batch Workloads TCM uses a multiple priority closed queueing model to predict task response time and transaction throughput for batch transactions. It is assumed that OLTP transactions have higher priority than batch transactions and that batch transactions do not interfere with OLTP activity. System resources that are not used for processing OLTP transactions are left for batch use.
Figure 28 Batch Applications Transaction Types TCM uses three transaction types: batch, online, and query. In TCM, a batch transaction is roughly equivalent to the common terms batch job and workload. The batch transaction can have multiple tasks. Batch transactions have the following characteristics: • Long job duration—Batch transactions can last several hours and can include processing of several tasks, each of which includes hundreds or thousands of transactions.
In contrast to the batch transaction or workload is the online transaction. The online transaction or workload is an open system. In contrast to batch transactions, the interval time of consecutive OLTP transactions is random. Although batch and online transactions are very different, query transactions are modeled as batch or online transactions, depending on the CPU demand of the query transaction. Large demand queries are modeled as batch. Small demand queries are modeled as OLTP.
However, if Mighty Manufacturing Company processed shipping orders in the morning and inventory analysis in the afternoon, these would be classified as two different batch transactions, each with one or more tasks. The purpose of defining multiple tasks for a batch transaction is so that TCM can estimate the response time and throughput per task. Differences in Modeling Batch Transactions When you model batch transactions in TCM, you need to modify your procedures in MeasTCM and TCM slightly.
Figure 29 Workload Apportionment Screen Notice that the batch transaction, CustUpdt, is labeled as B. The online transactions, Withdrawal and Deposit, are labeled as X. Number of Tasks In your batch transaction, you can have multiple tasks. In the Mighty Manufacturing example earlier, the Shipping Orders batch workload has two tasks. If your transaction has multiple tasks, you need to modify the #Tasks field for the batch transaction in the Workload Apportionment screen: 1. Open a WA model. 2.
• Calculate batch response times with completed jobs removed from system • CPU tunability The first three parameters, Number of Batch CPUs, and Hardware utilization thresholds for CPU and Disk, and Calculate Batch response times with completed jobs removed from system are found on the TCM Parameters screen. The last field is found on the Actual or Plan Baseline Detail by Category screen in the Performance model. The fourth is not a parameter but a column.
Figure 30 TCM Parameters Screen CPU Tunability The CPU Tunability field can be specified for each process category. It is used only for batch activity and is ignored for OLTP activity. It indicates whether a process category runs in one or all available batch CPUs. The default is N (no) for process categories that participate in batch transactions only. N (no) indicates batch transactions are assigned to run in only one CPU.
Figure 31 Plan Baseline Detail by Category Screen Consumption Model You can modify two batch-related fields in the Consumption model screen—#Txns per Task and Avg Batch Resp Time Constraint. Figure 32 (page 103) shows a sample Consumption model screen. Figure 32 Consumption Model Screen The batch response time is the time to perform one task. Thus, if you increase the #Txns per Task field, the (task) response time increases.
Modify the Avg Batch Resp Time Constraint field if you want TCM to estimate the number of CPUs and disks required to ensure that average batch response time does not exceed the constraint, and that all other constraints are also met. For example, if the average batch response time should not exceed five minutes, enter one of the following in this field: 0h5m 5m The syntax for the Avg Batch Resp Time is: [ number h ] [ number m ] [ number s] number is any nonnegative integer; for example, 0, 1, or 45.
To calibrate Batch Response Time: 1. Open a Performance model. 2. Select Update > Calibrate Batch Response Time. The Calibrate Batch Response Time dialog box appears. 3. Select the batch transaction name to calibrate from the list. 4. Enter the observed response time in the Calibrate Batch Response Time dialog box or accept the defaults. You can enter hours, minutes, and seconds. 5. Click OK. TCM adjusts its estimated response times in the Performance model by factors that reflect observed response times.
8 Workload Apportionment Model This chapter describes the screens and menus of the Workload Apportionment model. To move from one screen to the next, select the name of the screen you want to view from the Find menu. The WA model consists of the following major screens: • Category Demand Report—provides a summary of the performance data for process categories as summarized by MeasTCM on the NonStop host system. (See Chapter 3: “Using MeasTCM”.
Category Demand Report The Category Demand Report screen summarizes the performance data for process categories collected on the NonStop host system. To load the data, select Create a New WA Model from the TCM Initial menu or Load Category Data from the Update tab. The data on this screen is used as input for the Workload Apportionment component of the model.
Significant Columns 4. Cat Name—Modifiable by TCM administrators only Cat Name is the name of the process category defined in MeasTCM on the NonStop host system. (See “Process Category (PCAT) Statements” (page 53).) Up to 100 categories are supported by MeasTCM. 5. Type—Not directly modifiable Type is the type of a process category as defined in MeasTCM on the NonStop host system. The maximum number of categories is 100 (including Other and Error).
Figure 35 CPU Workload Apportionment Screen Significant Rows 1. CPU Config—Not directly modifiable CPU Config is the actual CPU configuration that was measured for the sample. The configuration includes the number of CPUs, type and the IPU information. For single core CPUs, the IPU field would contain "NA". If the sample was measured on an HP NonStop system running the G06 RVU, the suffix G06 is displayed next to the CPU type. The WA model permits only one CPU type to perform its calculations.
6. Second Unlabeled Column—Not directly modifiable The second unlabeled column indicates Aud if the transaction is audited but is left blank otherwise. A transaction is considered audited if it has a nonzero entry in the TMF process category in the Apportionment Weights table. 7. # Tasks—Modifiable by the TCM administrator # Tasks shows the number of tasks on the NonStop host system during the time period of the sample window. This number applies to batch transactions only. 8.
disk volumes, their configuration type, their write cache (WCE) configuration status and TMF audit trail count. Valid disk write cache configuration status can be: Enabled WCE enabled for disks that support WCE Disabled WCE disabled for disks that support WCE NA NA for disks that do not support WCE For audited disk configurations, valid TMF audit trails can be between 1 and 16. For unaudited configurations, it is NA (not applicable).
file. For more information, see “Transaction Count Table” (page 118) and the “Transaction Count Table Versus Transaction Count File” (page 120). 9. Read Seconds per Transaction—Not modifiable The apportioned number of disk read and disk write seconds consumed per transaction. The Average Txn column contains the (weighted) average of disk seconds for each transaction. Disk seconds are the number of seconds spent doing physical disk access, including seeks.
as type F if its CPU and disk consumption do not increase as throughput increases, such as during asynchronous data communications polling activity or the running of Measure. 3. Apportionment—Not directly modifiable Apportionment indicates the apportionment method used by the WA screen. The apportionment method name is the name of the apportionment table file (without the .AWA extension) that was used to generate the apportionment formulas when Workload Apportionment was selected from the Update menu.
screen is used as input for the TCM Performance model described in Chapter 6: “Creating a Performance Model”. Unlike the Workload Apportionment screens that show CPU, disk read, and disk write information on separate screens, the Workload Summary screen displays all the information on one screen. Figure 38 (page 114) shows a sample Workload Summary screen. Figure 38 Workload Summary Screen Significant Rows 1.
WA Model TCM Parameters Screen You can use the Workload Apportionment model TCM Parameters screen to change the default process priorities and to set the Disk Process Apportionment Factors, History File Name, and Default Process Priorities parameters for the WA model. Figure 39 (page 115) shows a sample TCM Parameters screen for a Workload Apportionment model. Figure 39 Workload Apportionment Model TCM Parameters Screen Significant Rows 1.
Significant Columns 4. Type—Not modifiable The Type column shows the process category types. 5. S Server process—this category type grows with throughput. R Requester process—this category type grows with throughput. D Disk process—this category holds all nonaudited disk processes. There is only one process with this category type. T TMF processes—this category holds all TMF-related processes, including the audit disk. TMF-related processes must be identified in the Process Category Control File.
Significant Row 1. User-Supplied Transaction Names—Not directly modifiable The name of the transaction as it appears on the Workload Apportionment screens. Transaction names are either initially defined on the NonStop host system using MeasTCM (see “TXN Entries” (page 59)) or named by default in TCM by server category (type S), but they can be overwritten on the Workload Apportionment screens by the TCM administrator. Significant Columns 2.
by selecting Calculate Now from the Options tab (if no table exists). For categories other than server, the default table assumes that category demand is spread evenly (on a per-transaction basis) across all transactions. For server categories whose names match a transaction name, the default table assumes that category demand is 1 for the transaction with the matching name and 0 for all others.
3. T TMF processes (including all audit disk processes) L Linehandler process O Other F Fixed (system overhead that does not vary with throughput) Z Ignore these categories (appropriate interrupt and disk activity to be omitted) Tunable CPU—Modifiable by the TCM administrator This column indicates, for batch transactions only, whether all process category activity occurs in one or multiple CPUs. It is ignored for online transactions.
For example, in a Pathway application, the default table would assume that the number of transactions is equal to the number of messages received by its Pathway server class. Although the default algorithm is frequently accurate, in some cases it will need to be modified. A value of zero in the matrix indicates that messages received (or sent) by the category should not be counted as transactions.
You cannot have any blank lines in the file. To enter comments, enter a comma following the transaction count number and type the comment. Using Process Pairs If a process category has process pairs, the process category is not a good choice to be used in the Transaction Count table because checkpoints to the backup process are included in the count for messages received.
Significant Row 1. Disk Process Priority Queuing—Modifiable by the TCM administrator Disk Process Priority Queuing shows whether disk process priority queuing is to be applied. Disk process priority queuing is also known as Mixed Workload Enhancement (MWE) and is provided for all applications running on NonStop systems using C30 or later product versions of the operating system.
Capacity History Screens The Capacity History screens–Database, TrendCharts, and MixCharts–are used to archive and organize apportioned performance data for a particular application or set of applications.
perform such operations as finding all samples where the throughput of transaction Txn 1 is greater than 10 transactions per second. 3. History Data—Not directly modifiable History Data is the historical sample data as apportioned by the WA model. This data is loaded by selecting Update History from the History tab and cannot be modified. Performance data is sorted by apportionment method and transaction type in descending date sequence. Significant Columns 4.
To select a worksheet either: • Click its sheet tab at the bottom of the screen • Use the TCM Find command to find it The TCM Find command can find the Trend charts or Mix charts (if present) as well as the History database. Trend Chart The TrendCharts worksheet has one Trend chart for each transaction name in the History data. Each Trend chart has two plot areas, one in the top half and one in the bottom half. The top half plots the CPU Seconds per Transaction.
Close (SF5) The Close option allows you to close all the WA model files, a selected apportionment table, or a Capacity History database. If you close the WA model, you return to the TCM Initial menu. Delete The Delete option allows you to delete any file on the workstation. If you select this option, it gives you a list of all the files in the current folder of the same type as the file you are viewing.
If you select Create Default Apportionment Tables, TCM assumes that the name of the apportionment tables you are working with is the same as the name of the apportionment method on the Workload Apportionment screen (with a file extension of .AWA). If this table name is not correct, enter the file name in the appropriate area of the dialog box. To apply these changes, click OK. If you select Change Current Apportionment Tables, a list of apportionment methods in the current folder appears.
from the database. This option is useful for deleting all history data for an old apportionment method. Sort History The Sort History option is used to sort the Capacity History data by transaction name or by sample name. It is available only when there is a Capacity History (.HST) file in the current session. To sort Capacity History data: 1. Select History > Sort History. 2. When the Sort History dialog box appears, click the button for the sort method. 3.
Print Tab The Print tab is used to print the screens of a model. Print (F8) The Print option prints one or more screens in the current model. A dialog box lists the names of the screens that you can choose to print and provides two print options.
Select Draft Quality to print the screen using your printer’s built-in character set. This option does not provide high-quality output, but it does offer fast printing. Select Preview to display an image of the printed page on the PC monitor before it is actually printed. This option is useful for checking that the page layout is correct for your particular printer configuration before you print the screen.
The scope of the Page Setup command is as follows: • Page Setup changes for a WA screen are applied to all WA model screens. • Page Setup changes for the Apportionment Weights table are applied to the Transactions Counts table and vice versa. • Page Setup changes for a sheet of the History workbook are applied only to that worksheet. Change Printer Fonts The Change Printer Fonts option changes the fonts that TCM uses to print its data.
Change Password The Change Password option changes the TCM passwords that control access to TCM administrator and superuser authorization levels. The default password for the TCM administrator level is “admin.” The password for the superuser level is “super.” To enhance TCM security, modify the default passwords. This option is available only to users with a TCM administrator or superuser authorization level.
Update History With Selected CSVs The Update History With Selected CSVs option enables you to update a History file with selected CSV files. If TCM Expert finds inconsistent transaction costs, it reports this information to you in a dialog box. Create Default Transaction Count Table The Create Default Transaction Count Table option enables you to create a default Transaction Count table based on a selected CSV file. It also checks for transaction cost consistency in the History file.
Pick Best CSV Sample Use the Pick Best CSV Sample option to select the most representative CSV file from all the available CSV files. When you select Pick Best CSV Sample, the Pick Best CSV Sample dialog box appears. Select a History file from the available files and click OK. TCM Expert computes the average CPU seconds for each transaction, assigns a weight to each transaction, and then picks the most representative sample across all samples. TCM Expert selects the best sample.
CPU Pathlength Threshold The CPU Pathlength Threshold field specifies the point at which a query transaction is considered either an OLTP or batch transaction. The default setting is two seconds. Small demand queries taking less than two seconds are considered OLTP transactions, and large demand queries taking longer than two seconds are considered batch transactions. Any TCM user can modify this factor.
9 The Performance Model The Performance model is used to answer what-if questions about baseline sample data that has already been computed in the Workload Apportionment model of TCM. You can analyze estimated throughputs, host response times, and the number of CPUs and logical disk volumes required to support different user-defined scenarios. The Performance model also provides presentation-quality charts for reporting estimated performance results.
◦ Disk write cache configuration ◦ Number of TMF audit trails required • TCM Parameters—used to set various parameters for the Performance model. • CPU Conversion Factors—used to set factors that represent the relative CPU speeds assumed for the application or set of applications in the model. • Baseline Detail by Category Actual—shows detailed information about process categories, CPU consumption by transaction, category priority, disk process priority queuing, and CPU tunability.
Screen Descriptions Help is available for fields on all the Performance model screens by selecting the field you want information about and selecting Help > TCM Help. The Read-only (protected) and User-modifiable (unprotected) fields of the screens have different background colors for easy identification. The default background of the Read-only fields is brown, and that of the User-modifiable fields is yellow.
Significant Rows 1. Application—Not directly Modifiable Application is the user-defined name of the application as it appears in the associated WA model. 2. Sample File—Not directly Modifiable Sample File is the file name of the associated WA model. 3. Sample Date—Not directly Modifiable Is the creation date of the input file which used to work with the performance model. 4. Planning Unit—Modifiable by the user Planning unit is the unit of transaction rate.
Significant Columns 13. Txn Name—Modifiable by TCM Administrator Txn Name is the name of the transactions as shown in the Workload Summary screen of the Baseline WA model. 14. Type—Not directly modifiable Type indicates the type of transaction. X indicates an online (OLTP) transaction, B indicates a batch transaction, and Q indicates a query transaction.
Figure 45 (page 141) shows a sample Consumption model. Figure 45 Consumption Model Significant Rows 1. Number—Constraint modifiable Number is the actual, constrained, and planned number of CPUs and disks required to achieve the performance specified by the constraints you set. The number of devices is for a well-tuned system. If the system is not well-tuned, more CPUs, disks, or both could be required. NOTE: 2. TCM does not model disk space.
model, the value of the Disk Write Cache is automatically set to the default write cache value corresponding to the selected Disk model. 6. TMF audit trails—Constraint Modifiable TMF audit trails depict the number of audit trails used in actual measurement. The value for the TMF audit trail constraint field can be between 1 and 16.
Significant Columns 13. Txn Name—Modifiable by TCM Administrator Txn Name is the name of the transactions as shown in the Workload Summary screen of the Baseline WA model. 14. Txn per Sec (or Hour, Day, or Month)—Plan modifiable Txn per Sec is the actual and planned transaction rates for each transaction in the model expressed in the user-defined planning unit (transactions per second, minute, hour, day, or month). The value in the Average Online Txn field shows the total transaction rate.
17. Disk Seconds—Modifiable through the Change Transaction option of the Update tab Disk Seconds is the actual and planned consumption, or cost, per transaction in disk seconds for each transaction. 18. Avg Resp Time—Not directly modifiable Avg Resp Time is the host response time for each transaction and the average or specified percentile transaction. This number includes both CPU and disk response times. It does not include communication line time and therefore is not end-to-end.
Figure 46 Planning Timeline Model CAUTION: Modifying a field erases the formula it originally contained. If you typed over data in a field, erasing the formula originally contained in the field, you can use the Refresh Formulas option on the Update tab to reinitialize the formulas in the screen. Significant Rows 1. Time Increment—Directly Modifiable Time Increment is the number of months between adjacent periods in the planning timeline. The initial default time increment is six months. 2.
5. # Disks, Disk Model, and Config—Directly Modifiable #Disks, Disk Model and Config are the desired number, Model, configuration of disks for any of the Scenario dates in the Planning Timeline model. Valid disk model numbers are listed in the Disk Factors screen. Valid disk configuration types are listed in the scratch areas of the TCM Parameters screen. Initially, the first column values match the actual number and model of disks, as shown in the first date column in the series.
12. #Txns—Not Directly Modifiable #Txns is the number of transactions per task if a batch transaction. This number can be trended using the Timeline Trends option of the Update menu. 13. Disk Partitions — Disk partitions are the actual, constrained, and planned partition configuration for SSD disks. Disk partition specifies if the SSD disk is partitioned or not. If partitioned, it also specifies the number of partitions. For non SSD disks, this field is NA (not applicable).
Figure 47 TCM Parameters Screen Significant Rows 1. Number of Batch CPUs—Directly Modifiable Number of Batch CPUs is the number of CPUs assigned to batch transactions. The default is All. For more information on this parameter, see “Number of Batch CPUs” (page 101). 2. Maximum CPU Utilization—Directly Modifiable Maximum CPU Utilization is the percentage of CPU that is available for batch activity. The default is 100 percent.
to FALSE, TCM assumes that all batch jobs restart when completed. No matter how long batch jobs actually take, they are all considered to be running simultaneously all of the time. The default value is TRUE. 5. Query Transaction OLTP-> Batch response Time CPU pathlength threshold—Directly Modifiable This option is the point at which a query transaction is considered either an OLTP or batch transaction. The default setting is two seconds.
Figure 48 CPU Factors Screen The leftmost column contains the names of the supported processor types in TCM. The first row contains the RVUs. The table contains the conversion factors for different configurations of CPU types and RVUs. The rightmost column contains the IPU count for multi-core processors and this field is not applicable for single core CPUs. When the CPU configuration types differ, the model's planned CPU consumption is estimated from the actual CPU consumption using these factors.
Figure 49 Disk Factors Screen Disk Factors Screen 151
The Performance Model
Disk Factors Screen 153
NOTE: Although RPM, read seek, and write seek times are mentioned in the disk factor sheet, they are not considered in the actual calculation of SSD disk values, as these three parameters are not applicable for SSD disks. Users can add a new disk in the blank rows at the end of the table. You must be either a TCM administrator or a super user to change the values in this screen. If the available fields are less than the number of new disks to be added, you can overwrite some of the unused existing entries.
The table is used to indicate how the CPU and disk seconds consumed by these categories are to be allocated to each transaction. 4. Type—Not Directly Modifiable Type is the type of a process category, as defined in MeasTCM on the NonStop host system. The category type consists of one of the following characters: 5.
Figure 51 Plan Baseline Detail by Category Screen Significant Rows 1. Txn Type—Not Directly Modifiable Online transactions are Txn Type X, batch transactions are Txn Type B, and query transactions are Txn Type Q. 2. Total CPU Secs—Not Directly Modifiable Total CPU Secs is the number of CPU seconds used by the transaction type. Significant Columns 3. Cat Name—Not Directly Modifiable Cat Name is the category names of all process categories as they appear on the Category Demand Report.
5. F Fixed (system overhead that does not vary with throughput) Z Ignore these categories (appropriate interrupt and disk activity to be omitted) Tunable CPU—Directly Modifiable Tunable CPU applies to batch transactions only. Y indicates batch transactions are spread evenly across all available batch CPUs. N indicates batch transactions are assigned to run in only one CPU. For more information, see “CPU Tunability” (page 102). 6.
NOTE: A given Capacity History screen is restricted to only one application (the name defined in the MeasTCM control file on the host) but can contain the results of more than one apportionment method. Figure 52 Capacity History Database Screen Significant Rows 1. Application—Not directly modifiable Application is the user-specified name of the application or set of applications that is being modeled. It is also the application name that appears on the Category Demand report of the WA model.
Capacity History Trend Charts Screen The TrendCharts sheet has one Trend chart for each transaction name in the History data. Each Trend chart has two plot areas, one in the top half and one in the bottom half. The top half plots the CPU Seconds per Transaction. The bottom half plots the TPS rate (if OLTP), or the Transactions per Task rate (if Batch) and a best-fit linear trend line. Sample dates are on the x-axis and apply to both halves of the chart.
Figure 53 TCM User Customization Screen Descriptions of five modifiable features of the TCM screens follow, along with the method to change them: • Font color of Manually overwritten cells When the formula of any user-modifiable cell is overwritten with a value, in either the Consumption Model, Planning Timeline model, CPU Factors screen, Disk Factors screen, or Detail Plan screen, the font of that cell becomes bold and takes on the color of the text Manual Overwrite Font ColorIndex in the User Customizat
The default value of this property is 5 (blue). Values less than 0 or greater than 56 are changed to the default value of 5. • Background color of Read Only cells The read-only protected cells of the TCM screens have the background color represented by the sample cell of the Read-Only cells feature of the User Customization screen. To change this feature, double-click the Read-Only cells label in the Feature column, select a color from the color palette that appears and click OK.
To customize a chart, select Customize from the Chart tab. A chart tools context-sensitive tab displays Excel’s chart formatting options for changing the chart type and text and graphic elements. You can customize the chart in a number of ways: • Double-click to view the chart formatting dialog, which allows you to set attributes such as borders, line styles, patterns, and colors. • Select any attached text and drag it to a new location on the chart.
Figure 55 Planning Timeline Model Chart for Batch Transaction Performance Model Ribbons The Performance model options are listed under tabs in the ribbon. Select tabs, options, and choices in dialog boxes in the same way as in the Work Apportionment (WA) model. File Tab The File tab is used to perform common operations on performance model files, such as opening or closing a model. Open (F4) The Open option opens a Capacity History database.
Exit TCM (SF5) The Exit TCM option ends the TCM session and returns you to Windows. Find Tab The Find tab, which you can select by pressing function key F7, lists the Performance model screens that you can display. If no charts are open in the current session, the Chart menu item is disabled to indicate that the option is not available. If the history database has not been opened, the Capacity History items are disabled.
Refresh Formulas The Refresh Formulas option reinitializes all the formulas or those you select through a dialog box. Use it to refresh the formulas if you typed over a lot of information while doing what-if exercises.
Timeline Trends The Timeline Trends option opens the history file for the application if it is not already in the current screen.
After the dialog box appears, choose the transaction to trend and optionally the start and end dates. Then click OK. The data is modified to show the trend in the Planning Timeline model. TCM uses a linear regression based on historical data to project the data. This option requires at least 10 Measure samples and a confidence level of 95 percent.
Merge Performance Models As a TCM administrator or superuser, you can merge two Performance models into one model or add transactions from one Performance model to another Performance model. These features are useful for modeling the impact of combining two applications in a single system. The models you are merging must have the same version and the same VPROC date. 1. Select Work with an Existing Model from the “TCM Initial Menu” (page 66) and click OK. The Open PM Model dialog box appears. 2.
TCM merges the Performance models and displays the Capacity Baseline–Plan of the merged model.
TCM merges the Performance models and displays the Capacity Baseline—Actual of the merged model.
Add Transaction The Add Transaction option adds a new, user-defined transaction to the Performance model that, typically, was not apportioned using a WA model. For example, if you have estimates of CPU and disk demands for a new transaction for which no Measure data yet exists, you can use this option to define the transaction. After you select this option, a dialog requests the name of the transaction to add.
The transaction’s activity is allocated entirely to a category of the same name and is added to both the actual and plan baseline screens. Add Similar Transaction The Add Similar Transaction option adds a transaction similar to an existing transaction in the model. Enter New Transaction Name and select one of the existing transactions in the model. Select Add. You must perform a Calculate in order for TCM screens to be properly updated with the newly added transactions.
Delete Transaction The Delete Transaction option deletes an existing transaction, and the category of the same name if you want, from a Performance model. After you select this option, a list of the transactions in the Performance model appears. To delete a transaction, select its name from the list and click Delete. Add Category The Add Category option adds a process category to the Performance model. Delete Category The Delete Category option deletes a category from the Performance model.
Widen Column The Widen Column option widens the selected column. Narrow Column The Narrow Column option narrows the selected column. Print Tab The Print tab is used to print the screens of a Performance model. Print (F6) The Print option prints one or more screens in the current model. A dialog box lists the names of the screens that you can choose to print and provides two print options.
Select Draft Quality to print the screen using your printer’s built-in character set. This option does not provide high-quality output, but it does offer fast printing. Select Preview to display an image of the printed page on the PC monitor before printing. This option is useful for checking that the page layout is correct for your particular printer’s configuration before you print the screen.
Change Printer Fonts The Change Printer Fonts option changes the fonts that TCM uses to print its data. Due to the difference between printers and monitors, different fonts are frequently required for printing and displaying data. TCM comes with a default printer font setting that might not be appropriate for your specific requirements. It is recommended that you select Preview from the Print option to ensure that your printer fonts are appropriate to your needs before printing a screen.
Change Password The Change Password option changes the TCM passwords that control access to TCM administrator and superuser authorization levels. The default password for the TCM administrator level is “admin.” The password for the superuser level is “super.” To enhance TCM security, the default passwords should be modified. This option is available only to users with a TCM administrator or superuser authorization level.
About The About option displays information about the TCM environment.
10 Using TCM Expert TCM Expert is a feature of TCM that enables you to do the following: • Automatically generate Apportionment Weights table from sample data • Update capacity history with selected samples • Create a default Transaction Count table • Pick the most representative sample • View and modify TCM Expert parameters TCM Expert provides a way to do capacity planning when you have limited knowledge of the application design or architecture.
of the CSV files selected and at that time tests all of the CSV samples for consistency. If TCM Expert finds inconsistencies, it reports this information in a dialog box. If you are a TCM super user, the Generate TCM Expert Apportionment Weights dialog box shows the percent of selected CSVs used to generate the data. If you are not a TCM super user, the dialog box does not show this field.
1. Select TCM Expert from the “TCM Initial Menu” (page 66). The Specify Custom Directory dialog box appears. 2. Specify the custom folder location of your CSV files. Highlight the custom folder of your CSV files and double-click the custom folder name. When you double-click the name, TCM updates this name on the custom folder field of the dialog box and also makes this the location of your working session. 3. Click OK to close the dialog box. You can now begin using TCM Expert.
5. 6. 7. Click OK. TCM Expert generates a default Transaction Count table and Apportionment Weights table and displays the newly generated “Transaction Count Table” (page 118). Examine the default Transaction Count table. If you want to make changes: a. Select Options > Unprotect Admin Fields. b. Enter the corrected values. c. Select Options > Protect Admin Fields. Examine the default Apportionment Weights table. To navigate to that table, select Find > Apportionment Weights Table.
From this dialog box, you can do the following: a. From the History Database list, select or enter the name of the Capacity History database that will be updated automatically after the weights are generated. The Capacity History database is created if it does not already exist. b. To select a name for the source of the transaction counts, select an apportionment table file. If you have transaction count files, TCM Expert automatically identifies them.
You can also select a different transaction or Transaction Count table from the lists in the dialog box. As you change the selected transaction or Transaction Count table, TCM updates the information in the dialog box to reflect your selections. When you finish entering changes, click OK to enter the changes and close the dialog box. 5. Once you complete the fields in the Generate Apportionment Weights dialog box, click Generate.
To override TCM Expert’s calculations for any category marked with a Y: a. Select Options > Unprotect Admin Fields. Modifiable fields are dimmed. b. Type an N in the Expert Flag column of the category you want to override and enter the values you believe are correct under the transaction columns. To finish entering your data, you must press Return before selecting Calculate Now or pressing a function key. c. Select Options > Protect Admin Fields. d.
7. 8. If you want, you can remove or add CSV files from the list to update capacity history: • To select CSV file, select a file name in the Unselected CSV Files field, and then click Add. The file name appears in the Selected CSV File list. • To clear a CSV file, click the file name in the Selected CSV File list and click Remove. The file name appears in the Unselected CSV File list.
name, TCM updates this name in the custom folder field of the dialog box and also makes this the location of your working session. 3. 4. Click OK to close the dialog box. Select TCM Expert > Pick Best CSV Sample. The Pick Best CSV Sample dialog box appears. 5. Select the Capacity History file from which to pick the CSV sample and click OK. TCM Expert picks the best CSV sample. 6. If you want to load this CSV sample into the Performance model, click Create.
A TCM File Names Distribution Folder for the PC Using Excel 2003 The TCM files are in Excel spreadsheet format unless otherwise stated: File Name Description 1TCM.MWA Excel macro sheet for the WA model. 1TCM.MXP Excel macro sheet for TCM Expert. DEMO.AWA Sample WA model for the Demo model. DEMO.HST Sample history file for the Demo model. DEMO2301.CSV Sample CSV file. EXAMPLE1.CSV Performance sample data (in Comma Separated Values format). EXAMPLE2.
Distribution Folder for the PC Using Excel 2007 The TCM files are in Excel spreadsheet format unless otherwise stated: File Name Description 1TCM.MWA Excel macro sheet for the WA model. 1TCM.MXP Excel macro sheet for TCM Expert. DEMO.AWA Sample WA model for the Demo model. DEMO.HST Sample history file for the Demo model. DEMO2301.CSV Sample CSV file. EXAMPLE1.CSV Performance sample data (in Comma Separated Values format). EXAMPLE2.
Custom Folder File Name Description application-name.HST Capacity history database for the modeled application or applications. custom.PXP An internal file, one per custom folder, that contains detailed diagnostic and results information beyond that displayed in the history file or the .AWA file. method-name.AWA WA apportionment tables for the modeled application or applications. sample-name.TCF Transaction count file for your application.
B MeasTCM Error Messages MeasTCM generates three types of error messages: • COBOL85 run-time diagnostic messages. For an explanation of these messages, see the COBOL85 Manual. • MeasTCM termination messages. These messages indicate normal or abnormal termination of MeasTCM. A list of these messages follows. The number preceding each message is the corresponding termination-info parameter. • Informational messages. These messages are for your information only and do not have message numbers.
Recovery Edit the MeasTCM control file and add an APPL entry defining the application name. 6 More than one CSVDATA entry in control file Cause The MeasTCM control file contains more than one CSVDATA entry. Effect The program terminates. No CSV files are produced. Recovery The MeasTCM control file can only contain one CSVDATA entry. Edit the control file and remove all but one CSVDATA entry. 7 More than one CSVLO entry in control file Cause The MeasTCM control file contains more than one CSVLO entry.
11 More than one APPL entry in control file Cause The MeasTCM control file contains more than one APPL entry. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and remove all but one APPL entry. 12 Missing APPL parameter in control file Cause The APPL keyword in the MeasTCM control file is not followed by an application name. Effect The program terminates. No CSV files are produced.
Process category number out of range n Cause One of the PCAT entries in the MeasTCM control file contains a process category number n that is larger than 97. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and correct the process category number. It must be an unsigned integer between 00 and 97. 19 Missing TXN parameter in control file Cause The MeasTCM control file contains one or more TXN entries without a corresponding transaction type and name.
Recovery Edit the MeasTCM control file and specify a valid time window using CSVLO (lower boundary) and CSVHI (upper boundary). 23 PCAT entry contains nonnumeric process category number text Cause One of the PCAT entries in the MeasTCM control file contains a nonnumeric process category number text. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and correct the process category number. It must be an unsigned integer between 00 and 97.
Cause The measurement data files contain CPU entity records showing mixed CPU types and the CONVERTTO entry is not present in the MeasTCM control file. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and add a CONVERTTO entry defining what the output CPU type should be. 28 No control file specified Cause No control file was specified at MeasTCM run time. Effect The program terminates. No CSV files are produced.
32 Duplicate Measure data file name Cause Two or more Measure entries in the MeasTCM control file specify the same Measure data file name. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and remove the duplicate Measure entries. 33 Invalid node name for process category n Cause One of the PCAT entries of process category n contains an invalid node (system) name specification. Effect The program terminates. No CSV files are produced.
37 Invalid program file name for process category n Cause One of the PCAT entries of process category n contains an invalid program file name specification. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and correct the program file name. It must be of the form volume.subvolume.filename. Wild-card characters asterisk (*) and question mark (?) are allowed.
Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and check that each process category has only one corresponding TITLE entry. 42 Invalid user ID user number for process category n Cause One of the PCAT entries of process category n contains an invalid user ID user number specification. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and correct the user ID user number.
Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and either add a TITLE entry for process category n or delete all the PCAT entries for that process category. 47 Missing CSVDATA parameter in control file Cause The CSVDATA keyword in the MeasTCM control file is not followed by a file name. Effect The program terminates. No CSV files are produced.
Cause Error n occurred during execution of the Measure callable procedure MEASGETVERSION. Effect The program abends. A process snapshot file is created. No CSV files are produced. Recovery For an explanation of the error codes, refer to the Measure Reference Manual. 52 Error during MEASREAD_DIFF_ n Cause Error n occurred during execution of the Measure callable procedure MEASREAD_DIFF_. Effect The program abends. A saveabend is created. Part of the CSV files might have been created but will be incomplete.
56 Internal error text Cause A MeasTCM internal error occurred. Effect The program abends. A saveabend is created. No CSV files are produced. Recovery Report this error to your HP analyst. 57 Too many Measure data files Cause Too many Measure data files are specified in the MeasTCM control file. Effect The program terminates. No CSV files are produced. Recovery “Features and Functions” (page 29) shows the maximum number of Measure data files that can be handled in one session.
61 Too many disk volumes Cause The total number of disk volumes defined in the Measure data files exceeds the maximum currently allowed by MeasTCM. Effect The program terminates. No CSV files are produced. Recovery Reduce the number of disk volumes by limiting the number of systems under study. If you cannot reduce the number of disk volumes, contact the GCSC. 62 Too many processes Cause The total number of processes in the current system exceeds the maximum currently allowed by MeasTCM.
Cause Error number n occurred during execution of the COBOLLIB routine COBOLASSIGN. If n = 1, the syntax for the file name specified in the REPORT entry of the MeasTCM control file was invalid. If n > 1, there was an internal MeasTCM error occurred. Effect The program terminates. No CSV files are produced. Recovery If n = 1, edit the MeasTCM control file and check that the REPORT entry refers to a valid local Spooler collector process name. If n > 1, report this error to your HP analyst.
Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and look for the TXN entry with the corresponding transaction type. Change the type parameter to a valid value. 70 Invalid CSV file name Cause The CSV data file name following the keyword CSVDATA in the MeasTCM control file is invalid. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and correct the CSV data file name.
Recovery Edit the MeasTCM control file and remove the erroneous CONVERTTO entry or entries. 75 Invalid CONVERTTO parameter in control file Cause The parameter following the CONVERTTO keyword in the MeasTCM control file does not represent a valid processor type. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and correct the CONVERTTO parameter.
Recovery Correct the value of CSVHI so that it is later than the start time of the first Measure data file. 81 CSVLO later than actual stop time Cause The user-specified CSVLO is later than the latest stop time across all Measure data files specified. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file so that the value of CSVLO is earlier than the latest stop time.
Essential CPU data missing. CPU=nn Cause The Measure data is incomplete. CPU number nn is missing. Effect The program terminates. No CSV files are produced. Recovery Rerun Measure. Include all CPUs (CPU*). 87 Entity overflow has occured. Cause A MEASURE entity overflow error occurred during the MeasTCM processing. Effect MeasTCM ignores the entities affected by the overflow. A count of the number of entities is kept and given in a separate message. The CSV files and audit report are produced if requested.
Recovery In the control file, alter the second parameter for the CONVERTTO factor. This parameter is a three-lettered entity. The list of allowable values is mentioned in the sample control file CNTLFILE. 91 Invalid DISKPOWER type parameter in control file. Cause In the MeasTCM control file, one or more than one of the DISKFACTOR keyword parameters is incorrect. Effect The program terminates. No CSV files are produced.
Recovery Edit the MeasTCM control file and add a DISKCONVERTTO entry defining the output disk type. 96 Measure data contains unsupported DISK type. Cause One of the Measure Disk entity records show a disk subtype of n, which this version of MeasTCM does not support. Effect The program terminates. A saveabend is created. No CSV files are produced. Recovery Check if this disk subtype is valid and obtain an updated product version of MeasTCM if necessary.
Recovery Edit the MeasTCM control file and correct the ACTUALDISK42 parameter. 101 Invalid ACTUALDISK43 parameter in control file. Cause The parameter following the ACTUALDISK43 keyword in the MeasTCM control file does not represent a valid disk of subtype 43. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and correct the ACTUALDISK43 parameter. 102 Invalid ACTUALDISK44 parameter in control file.
Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and add an ACTUALDISK44 entry defining the disks of subtype 44. 106 More than one ACTUALDISK53 entry in control file. Cause The MeasTCM control file contains more than one ACTUALDISK53 entry. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and remove the erroneous ACTUALDISK53 entry or entries. 107 Invalid ACTUALDISK53 parameter in control file.
Cause The MeasTCM control file contains more than one ACTUALDISK51 entry. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and remove the erroneous ACTUALDISK51 entry or entries. 111 Invalid ACTUALDISK51 parameter in control file. Cause The parameter following the ACTUALDISK51 keyword in the MeasTCM control file does not represent a valid disk of subtype 51. Effect The program terminates. No CSV files are produced.
Invalid or Blank IPU count parameter in CPUFACTOR Keyword. Cause The IPU count value (fifth parameter) for the CPUFACTOR keyword is not in the correct format; it is not numeric. Effect The program terminates. No CSV files are produced. Recovery Correct the value of the IPU count parameter for this keyword. 116 Invalid parameter for NUMTMFTRAIL Keyword. Cause The parameter following the NUMTMFTRAIL keyword in the MeasTCM control file does not contain a valid numeric between 1 and 16.
Recovery Edit the MeasTCM control file and add a DISKCONVERTTO entry with SAS disk entry (of subtype 51). 122 Disk Subtype is 52. ACTUALDISK52 keyword required. Cause The measurement data file or files contain disk entity records for disks of subtype 52, and the ACTUALDISK52 entry is not present in the MeasTCM control file. Effect The program terminates. No CSV files are produced. Recovery Edit the MeasTCM control file and add an ACTUALDISK52 entry defining the disks of subtype 52.
Recovery None Opening Measure data file measure file. Cause MeasTCM is opening the specified Measure file. Effect Informational only. Recovery None Closing Measure data file(s). Cause MeasTCM is closing all the Measure files it had opened. Effect Informational only. Recovery None Analyzing CPU records. Cause MeasTCM is processing CPU records. This message is displayed for each system found in the Measure files. Effect Informational only. Recovery None Writing CSV file csv file.
Effect Informational only. Recovery None Linking Discopen records. Cause MeasTCM is linking the process records to the discopen records in the Measure files. This message is displayed for each system found in the Measure files. Effect Informational only. Recovery None Sorting data for process category audit report. Cause MeasTCM is sorting the data for the requested audit report. This message appears only if an audit report was requested. Effect Informational only.
Warning: This Measure file does not overlap previous valid file. Cause This Measure data file was started after the previous file ended. There is no overlap. Effect The CSV files will not be 100 percent precise because there is a gap in the measurement window. Recovery If this situation is unacceptable, re-create Measure files by using intervals instead of Measure stops and starts. Collection interval not available; computed as stop time - start time.
Effect MeasTCM does not use this Measure file. Recovery This situation is unacceptable. Change the values for CSVHI, CSVLO or both values in the MeasTCM control file if the Measure file is to be included. Warning: This file will not be used to generate CSV file(s). Cause MeasTCM ignores this Measure file. Effect MeasTCM ignores this Measure file. Recovery A previous message indicates why the Measure file is not used. See the recovery for that message.
Recovery The Measure file is probably full (error 45). If the resulting CSV files are not acceptable, new Measure files are needed. Create larger Measure files, use fewer intervals, or use a smaller time window. Starting processing for CSV file. Cause MeasTCM is processing the data needed for the specified CSV file. Effect Informational only. Recovery None.
C TCM Menus This appendix shows the menu items for the TCM Workload Apportionment and Performance Model menus. You can print this appendix and use it as a job aid.
TCM Menus
For Performance Model Menu Items for Excel 2003 223
TCM Menus
Ribbon Items for Excel 2007/2010 For Workload Apportionment Model Ribbon Items for Excel 2007/2010 225
TCM Menus
For Performance Model Ribbon Items for Excel 2007/2010 227
TCM Menus
D TCM Response Time Formulas TCM uses these response time formulas for batch transactions: TCM uses these simplified response time formulas for online transactions: Query transactions are modeled as OLTP or batch, depending on the setting of the CPU Pathlength Threshold field.
E Common TCM Procedures This appendix gives step-by-step instructions for completing the following TCM procedures: • “Adding a Transaction” (page 231) • “Analyzing Capacity History” (page 231) • “Calculating Throughput Trending” (page 232) • “Calibrating Batch Response Time” (page 232) • “Changing Authorization Levels” (page 232) • “Changing Default Priorities” (page 233) • “Changing Expert Parameters” (page 233) • “Changing Individual Priorities” (page 234) • “Changing Passwords” (page 234
• “Returning to TCM From a Chart” (page 245) • “Running MeasTCM” (page 245) • “Saving Apportionment Tables” (page 246) • “Saving a Performance Model” (page 246) • “Saving a Workload Apportionment Model” (page 247) • “Sorting Capacity History Data” (page 247) • “Uncalibrating Batch Response Time” (page 248) • “Updating the Capacity History Database” (page 248) • “Updating the Capacity History Database With Selected CSV Files” (page 248) • “Using Online Help” (page 249) • “Using TCM Exper
1. 2. Open a Workload Apportionment model. Find the Capacity History database. If it is not accessible, open the desired history file. The Capacity History screen appears. 3. Analyze the historical variance of the CPU and disk demands for the sample data, ignoring or deleting unrepresentative data samples that might have been taken when data communications lines were inoperative or during CPU failures. 4.
a. b. 3. 4. Select Options > Change Authorization Level. The TCM Password dialog box appears. Click the User Authorization level button located at the top left corner. The TCM Password dialog box appears. At the prompt, enter the password for the new authorization level if one has been set up. Otherwise, enter the default password. The default password for the Administrator authorization level is ADMIN. The default password for the Super authorization level is SUPER.
6. Accept the default CPU Pathlength Threshold or specify a new one. The CPU Pathlength Threshold field specifies the point at which a query transaction is considered either an OLTP or batch transaction. The default setting is two seconds. Small demand queries taking less than two seconds are considered OLTP transactions, large demand queries taking longer than two seconds are considered batch transactions. Any TCM user can modify this factor. 7. Click OK. TCM Expert uses the new parameters.
To make sure a method is reasonable, save a model for each apportionment method and follow the steps in “Updating History With Several Examples” (page 82). Check the workload apportionment model of your application by comparing your results from all samples by reviewing the Capacity History file.
c. d. Click Update History and Display List of CSV Files, and then click Open. Select the CSV file or files to load and click OK. When you are finished, each data sample has a corresponding sample WA model in the folder. Creating a Mix Chart To create a Mix chart: 1. Open a Workload Apportionment model. 2. Select History > Update History. 3. Select History > Chart History.
3. Select History > Chart History. • If the History file contains data for more than one apportionment method, TCM prompts you to select the desired apportionment method to use when creating the History charts. • 4. If TrendCharts sheets, MixCharts sheets, or both are already part of the History file, the History Chart Overwrite Warning dialog box appears. To create new Trend charts and Mix charts (the existing ones are deleted automatically), click both checkboxes. This approach is recommended.
The sample cell immediately changes to the selected color. The default is yellow. To change the Screen Display font properties: 1. Double-click the Screen Display Fonts label in the Feature column. A dialog box appears. 2. Select one or more of the font types: Normal, Bold, Italic, or Title. Click OK. 3. Select the font properties from the Format Font dialog box that appears. The table is immediately updated to reflect the changes. To change the Screen Printer Font properties: 1.
Getting Help for Function Keys To get help for function keys either: • Select Help > Function Keys for help on commands available with the function keys. • Press the Shift key and the F1 function key simultaneously for a list of TCM function keys and their meanings.
Getting Help for Specific Fields TCM offers online, context-sensitive help messages that provide information on any field in the WA model or Performance model. To get help for specific fields: 1. Select the field about which you want more information. Either: 2. 3. 4. • Position the cursor on that field with the cursor keys. • Use the mouse to point to that field and click the left mouse button. Select help. Either: • Select Help > TCM Help. • Press the F1 function key.
3. 4. 5. Select a file from the list or enter the file name you want to open, including its .SWA or .SPM extension. Click OK. TCM loads the model file (rather than the Excel workspace file). Select File > Save to resave the model. Merging Performance Models As a TCM administrator or super user, you can also merge two Performance models into one model or add transactions from one Performance model to another Performance model.
5. Select Options > Unprotect Admin Fields. Cells whose values can be changed appear in blue. 6. 7. Enter all changes into the table. Select Options > Calculate Now (F12) to update the table. TCM recalculates values based on your changes. 8. Select Options > Protect Admin Fields. For more information, see “Apportionment Tables” (page 71). Modifying Apportioned Data Using an .
c. Select Options > Calculate Now (F12). TCM apportions the database according to the table you select. d. Select Options > Protect Admin Fields. 7. To save the modified .AWA file, select Save or Save As from the File tab. For more information, see “Apportionment Tables” (page 71). Modifying the Percentile Response Times TCM can provide the following results for OLTP transactions: average, 90th, 95th, or 99th percentile. The default is average.
6. When you are done, select Options > Protect Admin Fields to protect the apportionment tables. Although this step is optional, it is good practice because it safeguards your data from unintentional corruption. 7. Select Options > Calculate Now (F12) to update the table. 8. If you want to save the file, choose Save or Save As from the File tab. For more information, see “Modifying the Transaction Count Table” (page 77).
5. Select the capacity history file from which to select the CSV sample and click OK. TCM Expert picks the best CSV sample. 6. To load this CSV sample into the Performance model, click Create. TCM loads the CSV sample into the Performance model and displays the Capacity Baseline Plan for the CSV sample. You are now in the Performance model and have left TCM Expert. For more information, see Chapter 10: “Using TCM Expert”Section 10, Using TCM Expert. Printing a Chart TCM creates charts in color.
Saving Apportionment Tables If you do not save the apportionment tables, they are automatically saved when you save the entire model. However, it is recommended that you save the tables as soon as they are created. The Save As command allows you to change the name of the apportionment tables. The Save command does not. You will probably want to change the name of the tables if you are modifying the tables of an existing model rather than simply modifying the default tables of a new model.
1. Choose Save or Save As from the File tab. A dialog box containing a list of files in the Excel session appears. Type a new name (other than TCM.SPM) for the model. 2. To save the model in a location other than the TCM custom directory, click the Browse button next to the Performance Model name. Ensure that the check box next to the Performance Model name is checked. Click the Save button to save the Model under the different name.
1. 2. Select History > Sort History. When the Sort History dialog box appears, click the button next to the desired sort method: • To see how much a transaction varies by sample, select Sort by Transaction. • To see a chronological list of data, select Sort by Sample. 3. Click OK. You can click the Help button in the Sort History dialog box for online help about the Sort History dialog box.
8. 9. To clear a CSV file, select the file name in the Selected CSV File list and click Remove. The filename appears in the Unselected CSV File list. Click Update History. TCM Expert updates capacity history and shows the updated file. If TCM Expert finds inconsistent transaction costs in the capacity history, it reports this information to you in a dialog box. Using Online Help TCM offers online, context-sensitive help messages that provide information on any field in the WA model or Performance model.
c. d. Select the CSV sample data files to be used by selecting the files and clicking either the Remove or Add buttons. To select multiple files press the Shift or Control keys while clicking the file names. Press the Shift key to select a contiguous set of file or files. Use the Control key to add or remove a highlighted file. If you want to get information about a CSV file before you decide whether to use it, select the CSV file name and click the CSV Info button. The CSV Info dialog box appears.
1. 2. Open a Performance model. Select Consumption model. The Consumption model appears. Actual data based on the workload-apportioned Measure data appears in the Actual column. Default constraints appear in the Constraint column. The configuration recommended by TCM appears in the Plan column. 3. Enter constraints in the input fields in the Constraint column and other selected cells. Then press Return.
Viewing TCM Expert Parameters 1. Select TCM Expert from the “TCM Initial Menu” (page 66). The Specify Custom Directory dialog box opens. 2. 3. Specify a custom folder and click OK. Select Expert > TCM Expert Parameters. The TCM Expert parameters appear.
Glossary accelerated mode The operational environment in which Accelerator-generated RISC instructions execute. application A set of processes that execute an end-user business function on an HP NonStop system. These processes are assigned to categories, and the application is given a name in the MeasTCM control file. Apportionment Weights table A table that defines the way in which process category performance data should be apportioned to transactions.
custom folder A folder where the performance data samples, custom models, and sample models for the application or set of applications under study reside. It is recommended that a different custom folder be established for each set of applications under study. Disk Factors A screen used to view disk model number, rotational speed (rpm), and controller. distribution folder The folder where the TCM distribution files are located.
online transaction A workload that processes one transaction at a time, as opposed to a batch transaction, which processes several transactions at once. performance data sample Performance information collected on the NonStop host system for a particular application or set of applications to be modeled. Performance data is collected using Measure, summarized using MeasTCM, and then downloaded to the PC. TCM uses the raw performance data as input to its modeling process.
host system and download the file to the custom folder, or you can create the file directly on the PC. Transaction Count table A table that defines how TCM should calculate the number of transactions from process category performance data. The table specifies how to compute the number of transactions from the number of messages process categories that receive, send, or both. This table is created and modified only by users with a TCM administrator or superuser authorization level.
Index Symbols # Commits per Txn field, 149 # CPUs and CPU Type field, 145 # Disks and Disk Type field, 146 # Tasks field, 110, 111, 113 # Txns field, 110, 111, 113 #Txns field, 147 #Txns per Task field, 114, 140 .
sorting, 247 Capacity History Database, 137 Capacity History database, 83 updating, 80, 248 Capacity History Database screen, 157 Capacity History Mix Chart screen, 159 Capacity History MixChart, 137 Capacity History screen, 106, 123 Capacity History screens, 157 Capacity History Trend Charts screen, 159 Capacity History TrendChart, 137 capacity planners TCM user, 20 capacity planning process description, 21 Cat Name field, 108, 117, 118, 154, 156 category 98 (_OTHER) checking, 235 category 98 (other) contr
DISKCONVERTTO entry control file, 44 DISKFACTOR entry control file, 46 Disks field, 139 Display Scratch Areas, 140 E error messages, 191 EXECUTION-LOG parameter MeasTCM, 60 Expert Flag field, 117, 119 Expert option, 68 F features, 18 fields getting help, 240 file defaults, 59 file structure, 65 files CNTLFILE, 33 transaction count, 74 Fixed CPU Sec Actual/Plan field, 157 Fixed CPU Sec field, 155 Fixed CPU Seconds field, 109 Fixed process, 108 Fixed Read Sec field, 111 Fixed Write Sec field, 112 function k
how to access, 64 online models, 31 order of parameters process category statements, 57 OSS process category type, 59 OTHER category control file, 62 P passwords changing, 234 PCAT statements, 53 percentile response times modifying, 90, 243 performance data downloading, 75 loading, 75 loading samples, 76 Measure, 29 Performance Model About option, 178 Add Category option, 173 Add Similar Transaction option, 172 Add Transaction option, 171 Calculate Now option, 176 Change Authorization Level option, 176 Cha
example, 57 process category (PCAT) statements, 53 wild card, 56 Process Category Audit report checking accuracy of control file, 251 MeasTCM, 62 process category priority changing, 233 process category statements order of parameters, 57 process category type disk, 59 Measure, 59 OSS, 59 requester, 59 server, 59 SQLMP, 59 SQLMX, 59 system, 59 TITLE statements, 58 TMF, 59 process category types, 108, 116, 117 process flow diagram, 21 process flow diagrams guidelines, 33 process pairs, 121 processor speed, 19
T TACL, 29 TCM Expert option, 68 file structure, 65 getting started, 64 Initial Menu, 66 passwords, 69 security, 68 status, 64 user levels, 64 TCM description, 18 TCM Expert Cache Miss Factor field, 134 Check History Over Last field, 134 CPU Pathlength Threshold field, 135 data splitting, 132 functions, 179 generating an Apportionment Weights Table for CSV files, 249 getting started, 180 overriding calculations, 185 viewing and changing parameters, 187 viewing parameters, 252 Workload Apportionment Weights
create new, 67 creating, 99, 236 customizing, 71 data source, 106 defined, 19 load CSV file, 99 major steps to create, 75 options, 67 saving, 81 work with existing, 67 Workload Apportionment (WA) model screen Apportionment Weights Table, 106 Capacity History, 106 Category Demand Report, 106 TCM Parameters, 106 Transaction Count Table, 106 Workload Apportionment, 106 Workload Detail by Category, 106 Workload Summary, 106 Workload Apportionment ribbons About option, 135 Calculate Now option, 131 Change Author
