Customizing IBM Tivoli Workload Scheduler for z/OS V8.2 to Improve Performance

Book description

Scheduling is generally considered as the nucleus of the data center, because the orderly, reliable sequencing and management of process execution is an essential part of IT management.

IBM Tivoli Workload Scheduler for z/OS is the IBM strategic product used in many large-to-midsized customer environments, responsible for scheduling critical batch applications. Therefore, the performance of Tivoli Workload Scheduler for z/OS is one of the important factors that affect the overall satisfaction from the IT services for these companies.

This IBM Redbooks publication covers the techniques that can be used to improve performance of Tivoli Workload Scheduler for z/OS (including end-to-end scheduling).

There are many factors that might affect the performance of any subsystem. In this book, we confine ourselves to those things that are internal to Tivoli Workload Scheduler, or can be easily verified and modified, and are likely to apply to the majority of Tivoli Workload Scheduler customer sites.

Although this book is aimed at those very large installations with a batch load of 100,000 or more jobs per day, it will also be relevant to installations with a smaller batch workload who are suffering from a shrinking batch window, or those who are trying to maximize the throughput on their existing hardware, or both.

Please note that the additional material referenced in the text is not available from IBM.

Table of contents

  1. Notices
    1. Trademarks
  2. Preface
    1. The team that wrote this redbook
    2. Become a published author
    3. Comments welcome
  3. Chapter 1: Introduction
    1. Definition of Tivoli Workload Scheduler performance
    2. Factors affecting performance
    3. Measuring performance
    4. Our lab environment
    5. New functions related to performance
      1. Multiple first-level domain managers
      2. Improved SCRIPTLIB parser
      3. Check server status before Symphony file creation
      4. Improved joblog retrieval performance
  4. Chapter 2: Tivoli Workload Scheduler for z/OS subtask interaction
    1. How Tivoli Workload Scheduler for z/OS interacts with the current plan
      1. The event manager
      2. Initialization statements affecting event handling
      3. The general service task
      4. Initialization statements affecting the general service task
      5. The normal mode manager
      6. Initialization parameters that affect the normal mode manager
      7. The workstation analyzer
      8. Parameters that affect the workstation analyzer
      9. Controlling access to the current plan
      10. Balancing access to the current plan
    2. Tuning Tivoli Workload Scheduler for z/OS throughput
      1. Breakdown of the workstation analyzer task
      2. Improving workstation analyzer throughput
      3. Software solutions
      4. Find-a-winner algorithm
    3. A day in the life of a job
  5. Chapter 3: Optimizing Symphony file creation and distribution
    1. End-to-end processing and performance issues
    2. Symphony file creation and distribution
    3. Preliminary testing (50,000 or fewer FTA jobs)
      1. FTA tuning parameters (localopts)
      2. Script library as a PDS in LLA instead of a PDSE
      3. Centralized scripts
    4. Initial test results (250,000 FTA jobs)
    5. How to tune E2E for faster Symphony creation and distribution
      1. z/OS UNIX System Services tuning and data set placement
      2. UNIX System Services tuning
      3. zFS tuning
      4. TOPOLOGY parameters
      5. FTA tuning
      6. Centralized scripts
    6. Tuning results and recommendations
    7. Additional tuning changes (z/OS UNIX System Services)
      1. Using an empty EQQSCLIB
      2. Defining SHAREOPTION(1) for EQQSCPDS file
      3. Defining BUFFERSPACE for EQQSCPDS file
    8. Final tuning results for Symphony creation and distribution
    9. Recommendations based on tuning results
  6. Chapter 4: Optimizing the UNIX System Services environment
    1. UNIX System Services overview
      1. UNIX overview
      2. What people like about UNIX
      3. What people do not like about UNIX
      4. UNIX operating system
      5. UNIX file system
      6. MVS and UNIX functional comparison
      7. z/OS UNIX System Services fundamentals
      8. Address spaces
      9. What people like about z/OS UNIX
      10. What people do not like about z/OS UNIX
    2. z/OS UNIX performance tuning
    3. z/OS UNIX file systems
      1. Hierarchical File System (HFS)
      2. Network File System (NFS)
      3. Temporary File System (TFS)
      4. zSeries File System (zFS)
      5. zFS aggregates
      6. Installing a zFS
      7. Tuning zFS
    4. HFS and zFS comparison (1/2)
    5. HFS and zFS comparison (2/2)
  7. Chapter 5: Using Tivoli Workload Scheduler for z/OS effectively
    1. Prioritizing the batch flows
      1. Why do you need this?
      2. Latest start time
      3. Latest start time: Calculation
      4. Latest start time: Maintaining
      5. Latest start time: Extra uses
      6. Earliest start time
      7. Balancing system resources
      8. Workload Manager integration
      9. Input arrival time
      10. Exploit Tivoli Workload Scheduler for z/OS restart capabilities
    2. Designing your batch network
    3. Job streams in an end-to-end environment
    4. Moving JCL into the JS VSAM files
      1. Pre-staging JCL tests: Description
      2. Pre-staging JCL tests: Results tables
      3. Pre-staging JCL conclusions
    5. Recommendations
      1. Pre-stage JCL
      2. Optimize JCL fetch: LLA
      3. Optimize JCL fetch: Exits
      4. Best practices for tuning and usage of resources
      5. Implement EQQUX004
      6. Review your tracker and workstation setup
      7. Review initialization parameters
      8. Review your z/OS UNIX System Services and JES tuning
  8. Chapter 6: Data store considerations
    1. What is the data store?
    2. When to use the data store
    3. Data store performance evaluation
    4. Conclusions and recommendations
  9. Chapter 7: Optimizing Job Scheduling Console performance
    1. Factors affecting the Job Scheduling Console performance
    2. Applying the latest fixes
    3. Resource requirements
    4. Setting the refresh rate
    5. Setting the buffer size
    6. Minimize the JSC windows to force the garbage collector to work
    7. Number of open editors
    8. Number of open windows
    9. Applying filters and propagating to JSC users
    10. Java tuning
    11. Startup script
  10. Chapter 8: Troubleshooting
    1. E2E troubleshooting: Installation
      1. EQQISMKD
      2. EQQDDDEF
      3. EQQPCS05
      4. EQQPH35E message after applying or installing maintenance
    2. Security issues with E2E
      1. Duplicate UID
      2. E2E server user ID not eqqUID
      3. CP batch user ID not in eqqGID
      1. CPUTCPIP not same as nm port
      2. PORTNUMBER set to PORT reserved for another task
      3. PORTNUMBER set to PORT already in use
      4. TOPOLOGY and SERVOPTS PORTNUMBER set to same value
    4. E2E Symphony switch and distribution problems
      1. EQQPT52E cannot switch to new Symphony file
      2. CP batch job for E2E is run on wrong LPAR
      3. Changing the OPCMASTER that an FTA should use
      4. No valid Symphony file exists
      5. DM and FTAs alternate between linked and unlinked
    5. Other E2E problems
      1. Delay in Symphony current plan (SCP) processing
      2. E2E server started before TCP/IP initialized
      3. Jobs run at wrong time
      4. CPUTZ defaults to UTC due to invalid setting
      5. Domain manager (DM) file system full
      6. CP batch job starting before file formatting has completed
      7. EQQW086E in controller EQQMLOG
    6. OMVS limit problems
      1. MAXFILEPROC value set too low
      2. MAXPROCSYS value set too low
      3. MAXUIDS value set too low
    7. Other useful E2E-related information
      1. Restarting an E2E FTA from the distributed side
      2. Adding or removing an E2E FTA
      3. Reallocating the EQQTWSIN or EQQTWSOU file
      4. E2E server SYSMDUMP with Language Environment (LE)
    8. Troubleshooting the data store
    9. Where to find messages in UNIX System Services
    10. Where to find messages in an end-to-end environment
  11. Appendix A: Using the EQQUX000 and EQQUX002 exits
    1. The workstation analyzer (WSA) subtask
    2. Improving workstation analyzer throughput
    3. EQQUX000 and EQQUX002: Installation
    4. EQQUX000 and EQQUX002: Implementation
      1. JCL fetch from specific libraries
      2. Model JCL fetched from JCL libraries
      3. Model JCL fetched from storage
      4. Insertion of JCL cards before first EXEC or PROC statement
      5. Amend CLASS= on JOBCARD
      6. General usage notes
    5. How to define data store destinations on a per job basis with EQQUX002
  12. Appendix B: Gathering statistics
    1. Gathering statistics
    2. Changing statistics gathering and more
    3. Using the job tracking log data
  13. Appendix C: Additional material
    1. Locating the Web material
    2. Using the Web material
      1. System requirements for downloading the Web material
      2. How to use the Web material
  14. Abbreviations and acronyms
  15. Related publications
    1. IBM Redbooks
    2. Other publications
    3. Online resources
    4. How to get IBM Redbooks
    5. Help from IBM
  16. Index (1/3)
  17. Index (2/3)
  18. Index (3/3)
  19. Back cover

Product information

  • Title: Customizing IBM Tivoli Workload Scheduler for z/OS V8.2 to Improve Performance
  • Author(s): Vasfi Gucer, Anna Dawson, Art Eisenhour, Stefan Franke, Clive Kennedy, John Misinkavitch, Stephen Viola
  • Release date: November 2004
  • Publisher(s): IBM Redbooks
  • ISBN: None