Teradata Database. Resource Usage Macros and Tables

Transcription

1 Teradata Database Resource Usage Macros and Tables Release 13. B A October 211

2 The product or products described in this book are licensed products of Teradata Corporation or its affiliates. Teradata, Active Enterprise Intelligence, Applications Within, Aprimo, Aprimo Marketing Studio, Aster, BYNET, Claraview, DecisionCast, Gridscale, Managing the Business of Marketing, MyCommerce, Raising Intelligence, Smarter. Faster. Wins., SQL-MapReduce, Teradata Decision Experts, Teradata Labs Logo, Teradata Raising Intelligence Logo, Teradata Source Experts, WebAnalyst, and Xkoto are trademarks or registered trademarks of Teradata Corporation or its affiliates in the United States and other countries. Adaptec and SCSISelect are trademarks or registered trademarks of Adaptec, Inc. AMD Opteron and Opteron are trademarks of Advanced Micro Devices, Inc. EMC, PowerPath, SRDF, and Symmetrix are registered trademarks of EMC Corporation. GoldenGate is a trademark of Oracle. Hewlett-Packard and HP are registered trademarks of Hewlett-Packard Company. Intel, Pentium, and XEON are registered trademarks of Intel Corporation. IBM, CICS, RACF, Tivoli, and z/os are registered trademarks of International Business Machines Corporation. Linux is a registered trademark of Linus Torvalds. LSI is a registered trademark of LSI Corporation. Microsoft, Active Directory, Windows, Windows NT, and Windows Server are registered trademarks of Microsoft Corporation in the United States and other countries. NetVault is a trademark or registered trademark of Quest Software, Inc. in the United States and/or other countries. Novell and SUSE are registered trademarks of Novell, Inc., in the United States and other countries. Oracle, Java, and Solaris are registered trademarks of Oracle and/or its affiliates. QLogic and SANbox are trademarks or registered trademarks of QLogic Corporation. SAS and SAS/C are trademarks or registered trademarks of SAS Institute Inc. SPARC is a registered trademark of SPARC International, Inc. Symantec, NetBackup, and VERITAS are trademarks or registered trademarks of Symantec Corporation or its affiliates in the United States and other countries. Unicode is a registered trademark of Unicode, Inc. in the United States and other countries. UNIX is a registered trademark of The Open Group in the United States and other countries. Other product and company names mentioned herein may be the trademarks of their respective owners. THE INFORMATION CONTAINED IN THIS DOCUMENT IS PROVIDED ON AN AS-IS BASIS, WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING THE IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, OR NON-INFRINGEMENT. SOME JURISDICTIONS DO NOT ALLOW THE EXCLUSION OF IMPLIED WARRANTIES, SO THE ABOVE EXCLUSION MAY NOT APPLY TO YOU. IN NO EVENT WILL TERADATA CORPORATION BE LIABLE FOR ANY INDIRECT, DIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING LOST PROFITS OR LOST SAVINGS, EVEN IF EXPRESSLY ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. The information contained in this document may contain references or cross-references to features, functions, products, or services that are not announced or available in your country. Such references do not imply that Teradata Corporation intends to announce such features, functions, products, or services in your country. Please consult your local Teradata Corporation representative for those features, functions, products, or services available in your country. Information contained in this document may contain technical inaccuracies or typographical errors. Information may be changed or updated without notice. Teradata Corporation may also make improvements or changes in the products or services described in this information at any time without notice. To maintain the quality of our products and services, we would like your comments on the accuracy, clarity, organization, and value of this document. Please teradata-books@lists.teradata.com Any comments or materials (collectively referred to as Feedback ) sent to Teradata Corporation will be deemed non-confidential. Teradata Corporation will have no obligation of any kind with respect to Feedback and will be free to use, reproduce, disclose, exhibit, display, transform, create derivative works of, and distribute the Feedback and derivative works thereof without limitation on a royalty-free basis. Further, Teradata Corporation will be free to use any ideas, concepts, know-how, or techniques contained in such Feedback for any purpose whatsoever, including developing, manufacturing, or marketing products or services incorporating Feedback. Copyright by Teradata Corporation. All Rights Reserved.

3 Preface Purpose This book describes, and provides procedures for, Teradata Database resource usage data and macros. Audience This book is intended for system programmers, system administrators, and other database specialists responsible for administering or managing Teradata Database. Supported Software Release This book supports Teradata Database 13.. Prerequisites You should be familiar with basic computer technology, Teradata Database, and the system console environment. It will be helpful to review or reference the following books: Introduction to Teradata Workload Management API: PM/API and Open API Performance Management Resource Usage Macros and Tables 3

4 Preface Changes to This Book Changes to This Book Release Teradata Database 13. October 211 Teradata Database 13. March 21 Teradata Database 13. November 29 Description Removed the note and some text from the Channel Traffic Columns section in the ResUsageSpma Table chapter. Updated the NumSets description in Chapter 11 ResUsageSps Table on page 115. Added SpareTmon1, SpareTmon2, and SpareTmon3 to the ResUsageSvpr spare columns table. Updated the descriptions of the MSGWORKTHREE and MSGWORKELEVEN columns in the ResUsageSawt table. Corrected definition of AwtFlowControlled and AwtFlowCtlCnt in the ResUsageSpma table. 4 Resource Usage Macros and Tables

5 Preface Changes to This Book Release Teradata Database 13. April 29 Description The Vproc5 and VprocType5 fields are now valid on all platforms. ResUsageSpdsk table is available for use and reports detailed usage of pdisks. Updated descriptions of summary mode to clarify what is being reported. Updated descriptions for file system fields in the ResUsageSpma table. Fields that were once marked as invalid on all platforms that are now available for use on some or all platforms have been updated. QWaitTimeMax, QLengthMax, and ServiceTimeMax have been changed to gather type "track" (not "count") which means they no longer need to be divided by the log intervals to obtain the average. WorkTypeMax field in both the ResUsageSps and ResUsageAwt table now have the gather type track. Removed references to xschmon utility. It is no longer supported. Added new fields to ResUsageSps and ResUsageSpdsk. Added new flow control fields and Worktype fields to ResUsageSawt. Added new fields to ResUsageSvpr to support TDWM and DBQL. Added new PDE, FSYS, and Teradata Virtual Storage fields to ResUsageSps. Added new MI fields to ResUsageSawt and ResUsageSvpr. Updated or added logical device fields such as input and output traffic columns as well as response time columns to ResUsageSldv. Changed NodeType to CHAR(8) to accommodate new node types with longer names. The field PGId has a new data type of SMALLINT. Also, summary mode of the ResUsageSps table uses the triplet of the PGId, VprType, and PPId fields. Removed instances of gather type countshft because the only difference between countshft and count is how data collection is implemented. Clarified CPU normilization in SPMA table Marked all the Extent I/O Driver columns in the SVPR table as invalid. The SpareTmon field in every table tracks the Capacity on Demand (COD) value. ResUsageSps.WorkTypeInUse values now have a gather type count. The value must always be divided by the CollectIntervals value. ResUsageSps.WorkTypeMax values reports a maximum of sampled values and not the actual maximum of all inuse AWTs. Updated the definition of the ResGeneralInfo view and added the new ResSvprView view definition to Chapter 14. Resource Usage Macros and Tables 5

6 Preface Additional Information Release Teradata Database 12. September 27 Description Updated fields in the ResUsageSpma table and ResUsageSvpr table. Added two Teradata Active System Management tables and associated macros and views. Added the SET ACTIVELOGTABLE and SET SUMLOGTABLE database commands. Added new categories to the ResUsageSpma, ResUsageSvpr, and ResUsageIvpr tables. Added new fields to ResUsageSpma, ResUsageIpma, ResUsageSpdsk, and ResUsageSvpr tables. Changed the default of all OneNode macros to '1-1' and the format of NCPUs to 'ZZ9'. Additional Information URL Description Use the Teradata Information Products Publishing Library site to: View or download a manual: 1 Under Online Publications, select General Search. 2 Enter your search criteria and click Search. Download a documentation CD-ROM: 1 Under Online Publications, select General Search. 2 In the Title or Keyword field, enter CD-ROM, and click Search. Order printed manuals: Under Print & CD Publications, select How to Order. The Teradata home page provides links to numerous sources of information about Teradata. Links include: Executive reports, case studies of customer experiences with Teradata, and thought leadership Technical information, solutions, and expert advice Press releases, mentions and media resources Teradata Customer Education designs, develops and delivers education that builds skills and capabilities for our customers, enabling them to maximize their Teradata investment. To maintain the quality of our products and services, we would like your comments on the accuracy, clarity, organization, and value of this document. Please teradatabooks@lists.teradata.com 6 Resource Usage Macros and Tables

7 Preface References to Microsoft Windows and Linux References to Microsoft Windows and Linux This book refers to Microsoft Windows and Linux. For Teradata Database 13., these references mean: Windows is Microsoft Windows Server bit. Linux is SUSE Linux Enterprise Server 9 and SUSE Linux Enterprise Server 1. Resource Usage Macros and Tables 7

8 Preface References to Microsoft Windows and Linux 8 Resource Usage Macros and Tables

9 Table of Contents Preface Purpose Audience Supported Software Release Prerequisites Changes to This Book Additional Information References to Microsoft Windows and Linux Chapter 1: Introduction Benefits of Using Resource Usage Data Accessing Resource Usage Data Setting Up and Maintaining Resource Usage Data An Overview of Resource Usage Data Topics Covered Types of Data Gathering Resource Usage Data Data Collection Data Logging Using Resource Usage Macros Programs That Use Resource Usage Data PM/APIs and Equivalent Teradata Performance Monitor Options Chapter 2: Planning Your Resource Usage Data Enabling ResUsage Tables Tables Based on Needed Reports ResUsage Tables Setting Collection and Logging Rates Determining the Type of Logging Rate To Set Determining the Logging and Collection Rate Values Resource Usage Macros and Tables 9

10 Table of Contents Using Summary Mode Using Active Row Filter Mode Optimizing Resource Usage Logging The Cost of Logging Logging Cost Contributors Operational Methods Chapter 3: Resource Usage Procedures Enabling RSS Logging Using ctl and xctl Using Database Window General Macro Input Format Parameter Use for One-Node, Multiple-Node, All-Node, and Group Macros Using One-Node Macros Using ByGroup Macros Saving and Analyzing Data From Prior Releases Executing Macros Function EXECUTE MACRO Syntax Using DISABLE LOGONS To Stop Logging Purging Data Chapter 4: Resource Usage Tables Physical Table Naming Conventions Working with ResUsage Tables Relational Primary Index Inserting Rows to Resource Usage Tables Occasional Event Data Data Content of Tables: General Description Column Descriptions About the "Invalid Platform" Column About the "Gather Method" Column Column Names Ending In Sum Column Names Ending In Max Column Categories Summary Mode in Resource Usage Tables Resource Usage Macros and Tables

11 Table of Contents SECTION 1 Node Resource Usage Tables Chapter 5: ResUsageScpu Table Summary Mode Spare Columns Chapter 6: ResUsageSpma Table Spare Columns SECTION 2 Vproc Resource Usage Tables Chapter 7: ResUsageSawt Table Summary Mode Spare Columns Chapter 8: ResUsageShst Table Summary Mode Spare Columns Chapter 9: ResUsageSldv Table Summary Mode Spare Columns Chapter 1: ResUsageSpdsk Table Summary Mode Resource Usage Macros and Tables 11

12 Table of Contents Spare Columns Chapter 11: ResUsageSps Table Summary Mode Active Row Filter Mode Spare Columns Chapter 12: ResUsageSvdsk Table Summary Mode Spare Columns Chapter 13: ResUsageSvpr Table Summary Mode Spare Columns SECTION 3 Views and Macros Chapter 14: Resource Usage Views ResGeneralInfoView ResCPUUsageByAMPView ResCPUUsageByPEView ResShstGroupView ResSldvGroupView ResSvprView Chapter 15: Resource Usage Macros Macro Output Format ResAWT Macros Resource Usage Macros and Tables

13 Table of Contents ResAWT Sample Output ResAWTByAMP Sample Output ResAWTByNode Sample Output ResCPUByAMP Macros ResCPUByAMP Sample Output ResCPUByAMPOneNode Sample Output ResAmpCpuByGroup Sample Output Normalized Viewing of CPU Usage by AMP ResCPUByPE Macros ResCPUByPE Sample Output ResCPUByPEOneNode Sample Output ResPeCpuByGroup Sample Output Normalized Viewing of CPU Usage by PE ResCPUByNode Macros ResCPUByNode Sample Output ResCPUOneNode Sample Output ResCPUByGroup Sample Output ResHostByLink Macros ResHostByLink Sample Output ResHostOneNode Sample Output ResHostByGroup Sample Output ResLdvByNode Macros ResLdvByNode Sample Output ResLdvOneNode Sample Output ResLdvByGroup Sample Output ResPdskByNode Macros: Pdisk Device Traffic ResPdskByNode Sample Output ResPdskOneNode Sample Output ResPdskByGroup Sample Output ResMemMgmtByNode Macros ResMemMgmtByNode Sample Output ResMemMgmtOneNode Sample Output ResMemByGroup Sample Output ResNetByNode Macros ResNetByNode Sample Output ResNetOneNode Sample Output ResNetByGroup Sample Output ResNode Macros ResNode Sample Output ResOneNode Sample Output ResNodeByNode Sample Output ResNodeByGroup Sample Output Resource Usage Macros and Tables 13

14 Table of Contents ResPs Macros ResPsByNode Macro Sample Output ResPsByGroup Macro Sample Output ResVdskByNode Macros ResVdskByNode Sample Output ResVdskOneNode Sample Output ResVdskByGroup Sample Output SECTION 4 Appendixes Appendix A: How to Read Syntax Diagrams Syntax Diagram Conventions Appendix B: ResUsageIpma Table Spare Columns Appendix C: ResUsageIvpr Table Summary Mode Spare Columns Appendix D: Partition Assignments Partition Assignments Table Conventions Partition Assignment Listing Appendix E: System Activity Reporter Using Sar Comparing Sar and Resource Usage Data Resource Usage Macros and Tables

15 Table of Contents Glossary Index Resource Usage Macros and Tables 15

16 Table of Contents 16 Resource Usage Macros and Tables

17 CHAPTER 1 Introduction Resource usage, or ResUsage, is the collection and reporting of statistical information about the operation of your operating system and Teradata Database. Benefits of Using Resource Usage Data Resource usage data is useful for the following purposes: Measuring system performance Measuring component performance Assisting with on-site job scheduling Identifying potential performance impacts Planning installation, upgrade, and migration Analyzing performance degradation and improvement Identifying problems such as bottlenecks, parallel inefficiencies, down components, and congestion Accessing Resource Usage Data Resource usage data is stored in system tables and views in the DBC database. Macros installed with Teradata Database generate reports that display the data. There are several applications and utilities available for viewing resource usage data. See Teradata Manager User Guide for more information on viewing resource usage data with Teradata Manager using Teradata Performance Monitor (PMON). As with other database data, you can access resource usage data using SQL if you have the proper privileges. You can also write your own queries or macros on resource usage data. Setting Up and Maintaining Resource Usage Data You need to decide what resource usage data you want to collect and the level of detail you want it to cover. Resource Usage Macros and Tables 17

18 Chapter 1: Introduction An Overview of Resource Usage Data Related Topics This manual documents the resource usage data and settings for a variety of installation configurations and environments in Chapter 2: Planning Your Resource Usage Data. To implement the settings you decide on, see Chapter 3: Resource Usage Procedures. Once you set up your resource usage data collection the way you want, the only maintenance required is to purge old data regularly. See Purging Data on page 44. For additional information on performance analysis and system tuning, see the following books: Workload Management API: PM/API and Open API Performance Management Teradata Manager User Guide An Overview of Resource Usage Data Topics Covered The following table lists topics covered by resource usage data. Resource usage data covers BYNET traffic on a node client-to-server traffic CPU utilization data tracking storage device traffic pdisk device traffic vdisk device traffic Priority Scheduler information AMP Worker Task (AWT) information memory management activity summary information Which includes point-to-point messaging, broadcast messaging, and merge activities. data for each communication link. overhead, user service, and time of session execution. positions of sessions in locking queues. the number of reads/writes and amount of data transferred as seen from the storage driver. pdisk I/O, cylinder allocation, and migration statistics. all the cylinders allocated by an AMP (which can come from any pdisks in the clique). data by Performance Group (PG) from the Priority Scheduler and the ability to report resource usage data by Teradata Active System Management (Teradata ASM) workloads (WDs). AWT statistics. memory allocation, aging, paging, and swapping. all data collected for a node or vproc. 18 Resource Usage Macros and Tables

19 Chapter 1: Introduction Gathering Resource Usage Data Types of Data You can think of resource usage data in the following general categories. Category aggsam count track tmon Description Aggregated sample data tracks the average value of multiple samples of a countable item, such as average queue length. Given the number of samples that the aggregated value represents, the value can be divided by that number. Counted data tallies the number of times an event occurred, such as the number of disk reads or writes during a period of time. Tracked data gauges the current value of a countable item, such as queue lengths during a period of time. Time-monitored data reports the time that a resource was in use, such as processor time. Unless otherwise stated, all times reported in resource usage tables are in hundreths of seconds. Gathering Resource Usage Data Data Collection Resource usage data gathering is a two-phase process as follows: Data collection Data logging During the data collection phase the RSS gathers information from the operating system, Parallel Database Extensions (PDE), and Teradata Database. Data collection continues for a user-specified period of time, called the collection interval. Data Logging In the logging phase, which occurs at the end of each logging period and consists of one or more complete collect intervals, the RSS writes all gathered data to resource usage tables. Resource Usage Macros and Tables 19

20 Chapter 1: Introduction Gathering Resource Usage Data Data Collection Macros and Routines Gather Buffer Collect Interval Collect Buffer Work Buffer Summary Work Buffer Log Interval Log Buffer (Data to be written to disk) Summary Log Buffer ResUsage Write Queue ResUsage Tables ResUsage Reports 197E5 2 Resource Usage Macros and Tables

21 Chapter 1: Introduction Using Resource Usage Macros Using Resource Usage Macros Resource usage macros produce reports from data collected in the resource usage tables. You can use the reports to analyze key operational statistics and evaluate the performance of your system. Like other macros, resource usage macros consist of one or more Teradata SQL statements stored in Teradata Database and executed by a single EXECUTE statement. In addition to the name of the macro, the EXECUTE statement for resource usage macros can include parameters to specify the following: A specific single node A group of nodes Starting and ending dates and times Starting and ending nodes of a range of nodes Refer to Chapter 3: Resource Usage Procedures for more information on the resource usage macros, and SQL Quick Reference for details about how to use the EXECUTE statement. Programs That Use Resource Usage Data The resource usage data are not used by just the resource usage macros. Resource usage data can also be enabled from any of these programs: Teradata Manager Teradata Performance Monitor (client application running on Windows) Performance Monitor/Application Programing Interface (PM/API) PM/APIs and Equivalent Teradata Performance Monitor Options The following table lists the PM/APIs, specifically the System Performance Monitor and Production Control (PMPC) requests, and the equivalent Teradata Performance Monitor windows. If an error message is generated for the command, it is listed as well. Resource Usage Macros and Tables 21

22 Chapter 1: Introduction Programs That Use Resource Usage Data Request Equivalent Teradata Performance Monitor Options Selected from the Main Window Possible Error Message Explanation MONITOR PHYSICAL RESOURCE Resource Monitor Physical Resource Data Unavailable. The RSS Collection Rate is not currently enabled. The user requested Resource monitoring, but the RSS Collection Rate is not enabled. Remedy: Enable the RSS Collection Rate by setting it to a non-zero value and resubmit this request. For information on how to resolve this error message, see Messages. MONITOR PHYSICAL SUMMARY Summary Monitor Physical Resource MONITOR VIRTUAL RESOURCE Resource Monitor Virtual Resource Data Unavailable. The RSS Collection Rate is not currently enabled. The user requested Resource monitoring, but the RSS Collection Rate is not enabled. Remedy: Enable the RSS Collection Rate by setting it to a non-zero value and resubmit this request. For information on how to resolve this error message, see Messages. MONITOR VIRTUAL SUMMARY Summary Monitor Virtual Resource For more information on System PMPC requests and other PM/APIs, see Workload Management API: PM/API and Open API. 22 Resource Usage Macros and Tables

23 CHAPTER 2 Planning Your Resource Usage Data This chapter describes the following items: Enabling ResUsage Tables Setting collection and logging rates Using Summary Mode Using Active Row Filter Mode Optimizing resource usage logging Enabling ResUsage Tables The default collection and logging settings provide a good start for basic system monitoring. The default results in the ResUsageSpma (SPMA) table being logged every 1 minutes (6 seconds). The applications that use the collect buffer generally require more frequent data updates; therefore, the collection rate is typically adjusted to 6 seconds. PMON and other collect buffer based applications: Teradata Manager or PM/API, do not require the logging rate to be changed or any of the tables to be enabled (for logging) in order to access the RSS data they use. The ResUsageSpma table provides a high level summary of how the system is operating and contains summarized or key elements from most of the other tables. If you want to record detailed statistics covered by any of the ResUsage tables, then you should enable them for logging, along with specifying the largest logging period that will meet your needs. You should not log data that you do not have a planned need for since this does incur additional database system overhead and uses up additional database space. Naturally, the more tables you enable for logging and the shorter the logging period used, the more overhead the system will use. Tables Based on Needed Reports If you plan on using the report macros provided in Chapter 15: Resource Usage Macros, then you need to enable the associated table. The following table lists more information. Resource Usage Macros and Tables 23

24 Chapter 2: Planning Your Resource Usage Data Enabling ResUsage Tables For... a list of resource usage macros and their associated tables See... Determining the Type of Logging Rate To Set on page 27. instructions on setting ResUsage tables Enabling RSS Logging on page 33. instructions on using macros descriptions and examples of the macros General Macro Input Format on page 37 and Executing Macros on page 4. Chapter 15: Resource Usage Macros. ResUsage Tables ResUsage tables are divided into two groups: Node and Vproc. The following table describes these two groups and the tables within each group. It also provides guidance about which ones to enable. Table Name Covers When You Should Enable Node Resource Usage Tables The tables in this group are controlled by the RSS Collection Rate and Node logging rate. ResUsageScpu Statistics on the CPUs within the nodes. When the performance analysis suggests that the overall performance is limited or to check if a program is spinning in an infinite loop on an individual processor. For example, saturation of a particular CPU on each node or on a particular node while others are idle could indicate a task always uses that CPU. Also, you should enable when the system is first brought online to verify the following: That all CPUs are functioning on all nodes There is a good load balance among the CPUs ResUsageSpma System-wide node information provides a summary of overall system utilization incorporating the essential information from most of the other tables. Use the columns in ResUsageSpma to view BYNET utilization. Note: The BYNET can transmit and receive at the same time, resulting in 1% transmitting and 1% receiving values simultaneously. Another method of determining BYNET utilization and traffic is to use the blmstat tool. To provide an overall history of the system operation. 24 Resource Usage Macros and Tables

25 Chapter 2: Planning Your Resource Usage Data Setting Collection and Logging Rates Table Name Covers When You Should Enable ResUsageIpma System-wide node information, intended primarily for Teradata engineers. Generally, this table is not used at customer sites. Vproc Resource Usage Tables The tables in this group are controlled by the RSS Collection Rate and the Vproc logging rate. ResUsageSawt ResUsageShst ResUsageSldv Data specific to the AMP worker tasks (AWTs). Statistics on the host channels and LANs that communicate with Teradata Database. System-wide, logical device statistics collected from the storage driver. When you want to monitor the utilization of the AWT and determine if work is backing up because the AWTs are all being used. To determine details about the traffic over the IBM Host channels to determine if there is a bottleneck. To observe the balance of disk usage. The storage device statistics are often difficult to interpret with disk arrays attached due to multi-path access to disks. Note: Use the ResUsageSvdsk table first to observe general system disk utilization unless specifically debugging at a low level. ResUsageSpdsk Statistics collected from the pdisk device. To obtain detailed usage information about pdisks. ResUsageSps Data by Performance Group (PG) ID from the Priority Scheduler. When you need to track utilization by the query Workload Definition (WD) level. ResUsageSvdsk Statistics collected from the vdisk device. To view the details of the disk usage across the AMPs to look for hot AMPs or other skew issues. ResUsageSvpr ResUsageIvpr Data specific to each virtual processor and its file system. System-wide virtual processor information, intended primarily for Teradata engineers. To view details about the resources being used by each vproc on the system. This table is useful for looking for hot AMPS or PEs that may be CPU bound or throttled on other resources. Generally, this table is not used at customer sites. Setting Collection and Logging Rates The default for RSS Collection Rate and Node and Vproc Logging Rate is 6. However, the typical collection rate setting will be determined by the collect buffer applications, such as PMON, which normally sets the collection rate to 6 seconds. When you have decided what rates to set, see Chapter 3: Resource Usage Procedures for details on how to set the rates. Resource Usage Macros and Tables 25

26 Chapter 2: Planning Your Resource Usage Data Setting Collection and Logging Rates Collection Rate Logging Rate Collection rate controls the frequency (number of seconds) at which resource usage data is made available to programs that access the collect buffer data. The term collection interval is the same as collection rate. The RSS Collection Rate controls the collection of resource usage data from both the virtual processors and the nodes. That is, the RSS Collection Rate implies the rate for both vproc and node collection. Logging rate controls the frequency (number of seconds) at which resource usage data is logged to the resource usage tables. Resource usage logging means the writing of resource data as rows to one or more of the ResUsage database tables. The tables are named DBC.ResUsagexxxx, where xxxx is the name of the resource usage table (for example, Spma, Ipma, and so on) as listed in ResUsage Tables on page 24. The shorter the logging period, the more frequently data is logged, and the more disk space is used. There are two types of logging rates. This logging rate vproc logging rate node logging rate Controls the logging of resource usage data for virtual processors to the resource usage tables. nodes to the resource usage tables. The Node and Vproc logging rates must be integer multiples of the RSS Collection Rate. When the system is so busy that the ResUsage table logging gets backed up, RSS will automatically double the logging period which effectively summarizes the data by providing values for a time period twice that provided by the previous logging period. If you see the resource usage logging rates change without user intervention, this means that the database is busy. When no longer busy, the system resumes logging as before. Note: New events that may occur in the event logs during this double logging period do not represent fatal errors but are informational to indicate that the automatic operations of the RSS are attempting to maintain data logging. 26 Resource Usage Macros and Tables

27 Chapter 2: Planning Your Resource Usage Data Setting Collection and Logging Rates Determining the Type of Logging Rate To Set Use the following table to determine which type of logging rate to set. For this table And these macros... Use this logging rate... ResUsageScpu n/a node ResUsageSpma ResUsageSawt ResUsageShst ResUsageSldv ResUsageSpdsk ResUsageSps ResUsageSvdsk ResUsageSvpr ResCPUByGroup ResCPUByNode ResCPUOneNode ResMemByGroup ResMemMgmtByNode ResMemMgmtOneNode ResNetByGroup ResNetByNode ResNetOneNode ResNode ResOneNode ResNodeByGroup ResNodeByNode ResAWT ResAWTByAMP ResAWTByNode ResHostByGroup ResHostByLink ResHostOneNode ResLdvByGroup ResLdvByNode ResLdvOneNode ResPdskOneNode ResPdskByNode ResPdskByGroup ResPsByNode ResPsByGroup ResPsByNodeWDJoin ResPsCPUByNodeWDJoin ResPsCPUWDJoin ResVdskOneNode ResVdskByNode ResVdskByGroup ResAmpCPUByGroup ResCPUByAMP ResCPUByAMPOneNode ResCPUByPE ResCPUByPEOneNode ResPeCpuByGroup node vproc vproc vproc vproc vproc vproc vproc Resource Usage Macros and Tables 27

28 Chapter 2: Planning Your Resource Usage Data Setting Collection and Logging Rates Determining the Logging and Collection Rate Values The system imposes two rules on logging and collection rates: 1 Intervals must evenly divide into 36 (the number of seconds in an hour). The following table shows valid collection and logging rates. The white area of the table shows rates recommended only for short-term use for debugging a specific issue. The highlighted area of the table shows rates recommended for production processing A practical log interval minimum during production processing is 6 seconds. Intermediate log intervals, such as 12 seconds or 3 seconds can also be used. The default rate is 6 seconds. If the system becomes very busy, it will automatically double the logging period. This effectively summarizes the data by providing values for a time period twice that of the previous logging period. The system automatically returns to logging back to the rate you set when it is no longer busy. 2 The collection and logging rates that support the resource usage macro that you want to run must both be greater than zero. For example, if you want to set the RSS Collection rate to 1, then the logging rate of the node/vproc could be 1 or any legal multiple as listed in the previous chart (for example, the next value in the series not listed is 5, which is not legal). If the collection and logging rates that you enter do not comply with these rules, the system displays a diagnostic error message, but does not update the rates. The message suggests rates that do comply with these rules and are close to those you entered. Note: Node and Vproc logging rates must be integer multiples of the RSS Collection Rate. Rates and enabled tables may be changed at any time and the changes take effect immediately. 28 Resource Usage Macros and Tables

29 Chapter 2: Planning Your Resource Usage Data Using Summary Mode Using Summary Mode You can use Summary Mode to reduce the system overhead from logging tables that produce multiple rows per logging period. Summary Mode helps reduce overhead by combining data from multiple rows into one or more summary rows based on specific criteria for each table. For example, if you want to log information provided in the ResUsageSvpr table but do not need data for each individual vproc, then use summary mode to produce one row per vproc type instead of one row per vproc. The ResUsageSpma table, in comparison, provides node level summary of key fields from most of the other ResUsage tables. When more details are required than the ResUsageSpma table provides then the next level of information is provided by using summary mode logging for the table of interest. This helps minimize the cost of the data logging. You can select summary mode for each table individually. See the description for each table for details on how summary mode affects that particular table. The general rule applied to the data in summary mode is that all the "Statistics Columns" data fields are added together (Summed) for the rows that are combined into a single summary mode row. (Statistics Columns along with the other different column categories are discussed in Column Descriptions on page 46.) For example, for the ResUsageSvpr table in summary mode, all the individual vproc rows of the same vproc "type" are combined into a single row. Since the data values are added together, you need to divide by the number of rows that made up the summary mode row. For example, divide by the number of AMPs on that node to determine the values for the AMP summary row data values, and by the number of PEs on that node to determine the values for the PE summary row data values. (To determine the number of AMP, PEs, and all other vproc types on your system, you can use the ResUsageSpma table or use the Vproc Manager utility.) Fields that represent a maximum statistic are not summed together. Instead the maximum value from the rows is used. For example, the ResUsageSvpr table MsgWorkQLenMax field in the summary mode row for the AMPs will contain the maximum value from all the AMP rows that would have been logged in non-summary mode. The fields that represent a minimum statistic are summarized by storing the minimum value from all the constituent rows. Summary mode has either no effect on the values of the Housekeeping Columns or it is specifically detailed in the description of each affected field. To enable Summary Mode, see Enabling RSS Logging on page 33. For more information on Summary Mode, see Summary Mode in Resource Usage Tables on page 51. Using Active Row Filter Mode Active Row Filter Mode reduces the overhead of logging for some of the ResUsage table by limiting the data rows that are logged to the database. For the tables for which this option is Resource Usage Macros and Tables 29

30 Chapter 2: Planning Your Resource Usage Data Using Active Row Filter Mode enabled, only those rows whose data have been modified during the current logging period will be logged. Active Row Filter mode is enabled by default on the following tables: ResUsageSawt ResUsageSpdsk ResUsageSvdsk ResUsageSps When Active Row Filter Mode is enabled on a table, it may appear that rows are missing when looking at the query results. This is because the index values of the inactive rows varies over time so that a row with one index may be logged one period but not in another. To determine if rows are not being logged to the database, you should look in the event logs for messages indicating that rows have been lost. For more information, see the RSS Active Row Filter Mode Enable setting description in the table for Using ctl and xctl on page 33. Note: For tables that have both active row filter mode and summary mode enabled, the active row filtering is applied after the summarization. For data fields that have persistent data across logging periods, the summarized rows may combine the data from both active and inactive rows. In these cases, the active row filtering reports summarized rows that have at least one active row contribution. Extra care must be given when evaluating data where active row filtering and Summary Mode are used together. For details, see the discussion in Summary Mode on page 131 of the chapter on the ResUsageSps table. 3 Resource Usage Macros and Tables

31 Chapter 2: Planning Your Resource Usage Data Optimizing Resource Usage Logging Optimizing Resource Usage Logging The Cost of Logging Logging Cost Contributors Operational Methods Logging resource usage data to database tables incurs costs: Writing to the database adds to the system I/O load. On a heavily loaded system, this could affect the production workload throughput. The rows written to the database take up space. If this space is never reclaimed, it will eventually grow to consume all available space in user DBC. In an extremely loaded system, it is possible that the RSS can fall behind in writing data to the database. Although it caches such data and will eventually catch up if given a chance, the RSS will be forced to start discarding rows if the system load persists and its cache capacity has been exceeded. Logging costs are difficult to quantify. They depend on a number of interrelated factors: How busy is the system Which resource usage tables are enabled What resource usage logging rates are in effect The system configuration (vproc, CPU, host driver, logical devices or device controllers) Use the following methods to optimize performance and reduce the cost of resource usage logging on your system: 1 Use Summary Mode to reduce the number of rows inserted into the resource usage tables if Summary Mode data provides sufficient information for your needs. Note: If resource usage logging terminates due to a lack of table space: a Delete rows from the appropriate table or make more space for it in USER DBC. b Restart resource usage logging by entering the appropriate SET RESOURCE command. 2 For tables with a large number of rows (for example, ResUsageSps), use Active Row Filter Mode to limit the number of rows written to the database each logging period and to minimize the amount of system resources used. 3 Avoid unnecessarily using or exhausting available disk space by doing the following: Never enable logging on tables that you do not intend to use. For example, logging only to the ResUsageSpma table provides a lot of useful information with a minimal operational load on the system. Use the largest rates that provide enough detail information for your purposes. Generally, you should use a logging rate no smaller than 6. The default rate is 6. Resource Usage Macros and Tables 31

32 Chapter 2: Planning Your Resource Usage Data Optimizing Resource Usage Logging These values can be adjusted any time, regardless of whether the database system is busy. New values take effect as soon as the adjustment command is issued. (For example, with ctl, when you issue the WRITE command.) 4 Purge old data from the ResUsage tables periodically. Related Topics For instructions on... enabling resource usage tables, setting collection and logging rates, and summarizing or filtering rows See... Enabling RSS Logging on page 33. purging old data from ResUsage tables Purging Data on page Resource Usage Macros and Tables

33 CHAPTER 3 Resource Usage Procedures This chapter describes the following resource usage macro topics: Enabling RSS logging General macro input format Executing macros (syntax) Enabling and disabling LOGONS and the affects on resource usage logging Purging old data Enabling RSS Logging By using one of the following interfaces you can enable tables, set collection and logging rates, and optionally summarize or filter rows. If you are running... You can enable logging by running... For instructions, see... Windows or Linux MP-RAS the ctl utility from the Teradata Command Prompt. the xctl utility in non-windowing or windowing mode from the command line. Using ctl and xctl on page 33. MP-RAS, Windows, or Linux Database Window (DBW). Using Database Window on page 35. Using ctl and xctl Before you set the ResUsage tables, determine which tables and controlling rates apply to the resource usage macros you want to run. For more information, see the following topics: Enabling ResUsage Tables on page 23. Setting Collection and Logging Rates on page 25. The Control GDO Editor utilities (ctl on Windows and Linux, and xctl on MP-RAS) are used to set various Teradata Database configuration settings. The RSS-related settings are presented on the RSS screen of these utilities. For detailed information on starting these utilities and modifying the settings, see the ctl and xctl chapters in Utilities. Resource Usage Macros and Tables 33