You are previewing IBM System Storage DS8000 Performance Monitoring and Tuning.
O'Reilly logo
IBM System Storage DS8000 Performance Monitoring and Tuning

Book Description

This IBM® Redbooks® publication provides guidance about how to configure, monitor, and manage your IBM DS8880 storage systems to achieve optimum performance, and it also covers the IBM DS8870 storage system. It describes the DS8880 performance features and characteristics, including hardware-related performance features, synergy items for certain operating systems, and other functions, such as IBM Easy Tier® and the DS8000® I/O Priority Manager.

The book also describes specific performance considerations that apply to particular host environments, including database applications.

This book also outlines the various tools that are available for monitoring and measuring I/O performance for different server environments, and it describes how to monitor the performance of the entire DS8000 storage system.

This book is intended for individuals who want to maximize the performance of their DS8880 and DS8870 storage systems and investigate the planning and monitoring tools that are available. The IBM DS8880 storage system features, as described in this book, are available for the DS8880 model family with R8.0 release bundles (Licensed Machine Code (LMC) level 7.8.0).

Table of Contents

  1. Front cover
  2. Notices
    1. Trademarks
  3. IBM Redbooks promotions
  4. Preface
    1. Authors
    2. Now you can become a published author, too!
    3. Comments welcome
    4. Stay connected to IBM Redbooks
  5. Part 1 IBM System Storage DS8000 performance considerations
  6. Chapter 1. IBM System Storage DS8880 family characteristics
    1. 1.1 The storage server challenge
      1. 1.1.1 Performance numbers
      2. 1.1.2 Recommendations and rules
      3. 1.1.3 Modeling your workload
      4. 1.1.4 Allocating hardware components to workloads
    2. 1.2 Meeting the challenge: The DS8000 product family
      1. 1.2.1 DS8880 family models and characteristics
    3. 1.3 DS8000 performance characteristics overview
      1. 1.3.1 Advanced caching techniques
      2. 1.3.2 IBM Subsystem Device Driver
      3. 1.3.3 Performance characteristics for z Systems
      4. 1.3.4 Easy Tier
      5. 1.3.5 I/O Priority Manager
  7. Chapter 2. Hardware configuration
    1. 2.1 Storage system
    2. 2.2 Processor memory and cache
      1. 2.2.1 Cache and I/O operations
      2. 2.2.2 Determining the correct amount of cache storage
    3. 2.3 I/O enclosures and the PCIe infrastructure
    4. 2.4 Disk subsystem
      1. 2.4.1 Device adapters
      2. 2.4.2 DS8000 Fibre Channel switched interconnection at the back end
      3. 2.4.3 Disk enclosures
      4. 2.4.4 DDMs
      5. 2.4.5 Enterprise drives compared to Nearline drives and flash
      6. 2.4.6 Installation order
    5. 2.5 Host adapters
      1. 2.5.1 Fibre Channel and FICON host adapters
      2. 2.5.2 Multiple paths to Open Systems servers
      3. 2.5.3 Multiple paths to z Systems servers
      4. 2.5.4 Spreading host attachments
    6. 2.6 Tools to aid in hardware planning
      1. 2.6.1 White papers
      2. 2.6.2 Disk Magic
      3. 2.6.3 Capacity Magic
  8. Chapter 3. Logical configuration concepts and terminology
    1. 3.1 RAID levels and spares
      1. 3.1.1 RAID 5 overview
      2. 3.1.2 RAID 6 overview
      3. 3.1.3 RAID 10 overview
      4. 3.1.4 Array across loops
      5. 3.1.5 Spare creation
    2. 3.2 The abstraction layers for logical configuration
      1. 3.2.1 Array sites
      2. 3.2.2 Arrays
      3. 3.2.3 Ranks
      4. 3.2.4 Extent pools
      5. 3.2.5 Logical volumes
      6. 3.2.6 Space-efficient volumes
      7. 3.2.7 Extent allocation methods
      8. 3.2.8 Allocating, deleting, and modifying LUNs and CKD volumes
      9. 3.2.9 Logical subsystem
      10. 3.2.10 Address groups
      11. 3.2.11 Volume access
      12. 3.2.12 Summary of the DS8000 logical configuration hierarchy
    3. 3.3 Data placement on ranks and extent pools
      1. 3.3.1 Rank and array considerations
      2. 3.3.2 Extent pool considerations
      3. 3.3.3 Easy Tier considerations
  9. Chapter 4. Logical configuration performance considerations
    1. 4.1 Reviewing the tiered storage concepts and Easy Tier
      1. 4.1.1 Tiered storage concept
      2. 4.1.2 IBM System Storage Easy Tier
    2. 4.2 Configuration principles for optimal performance
      1. 4.2.1 Workload isolation
      2. 4.2.2 Workload resource-sharing
      3. 4.2.3 Workload spreading
      4. 4.2.4 Using workload isolation, resource-sharing, and spreading
    3. 4.3 Analyzing application workload characteristics
      1. 4.3.1 Determining skew and storage requirements for Easy Tier
      2. 4.3.2 Determining isolation requirements
      3. 4.3.3 Reviewing remaining workloads for feasibility of resource-sharing
    4. 4.4 Planning allocation of disk and host connection capacity
    5. 4.5 Planning volume and host connection spreading
      1. 4.5.1 Spreading volumes for isolated and resource-sharing workloads
      2. 4.5.2 Spreading host connections for isolated and resource-sharing workloads
    6. 4.6 Planning array sites
    7. 4.7 Planning RAID arrays and ranks
    8. 4.8 Planning extent pools
      1. 4.8.1 Overview of DS8000 extent pool configurations with Easy Tier
      2. 4.8.2 Single-tier extent pools
      3. 4.8.3 Multitier extent pools
      4. 4.8.4 Additional implementation considerations for multitier extent pools
      5. 4.8.5 Extent allocation in homogeneous multi-rank extent pools
      6. 4.8.6 Balancing workload across available resources
      7. 4.8.7 Extent pool configuration examples
      8. 4.8.8 Assigning workloads to extent pools
    9. 4.9 Planning address groups, LSSs, volume IDs, and CKD PAVs
      1. 4.9.1 Volume configuration scheme by using application-related LSS/LCU IDs
    10. 4.10 I/O port IDs, host attachments, and volume groups
      1. 4.10.1 I/O port planning considerations
    11. 4.11 Implementing and documenting a DS8000 logical configuration
  10. Part 2 IBM System Storage DS8000 performance management
  11. Chapter 5. Understanding your workload
    1. 5.1 General workload types
      1. 5.1.1 Typical online transaction processing workload
      2. 5.1.2 Microsoft Exchange Server workload
      3. 5.1.3 Sequential workload
      4. 5.1.4 Batch jobs workload
      5. 5.1.5 Sort jobs workload
      6. 5.1.6 Read-intensive cache friendly and unfriendly workloads
    2. 5.2 Database workload
      1. 5.2.1 Database query workload
      2. 5.2.2 Database logging workload
      3. 5.2.3 Database transaction environment workload
      4. 5.2.4 Database utilities workload
    3. 5.3 Application workload
      1. 5.3.1 General file serving
      2. 5.3.2 Online transaction processing
      3. 5.3.3 Data mining
      4. 5.3.4 Video on demand
      5. 5.3.5 Data warehousing
      6. 5.3.6 Engineering and scientific applications
      7. 5.3.7 Digital video editing
    4. 5.4 Profiling workloads in the design phase
    5. 5.5 Understanding your workload type
      1. 5.5.1 Monitoring the DS8000 workload
      2. 5.5.2 Monitoring the host workload
      3. 5.5.3 Modeling the workload and sizing the system
  12. Chapter 6. Performance planning tools
    1. 6.1 Disk Magic
      1. 6.1.1 The need for performance planning and modeling tools
      2. 6.1.2 Overview and characteristics
      3. 6.1.3 Disk Magic data collection for a z/OS environment
      4. 6.1.4 Disk Magic data collection for Open Systems environment
      5. 6.1.5 Configuration options
      6. 6.1.6 Workload growth projection
      7. 6.1.7 Disk Magic modeling
    2. 6.2 Disk Magic for z Systems
      1. 6.2.1 Processing the DMC file
      2. 6.2.2 z Systems model to merge the two DS8870 storage systems to a DS8886 storage system
      3. 6.2.3 Disk Magic performance projection for the z Systems model
      4. 6.2.4 Workload growth projection for a z Systems model
    3. 6.3 Disk Magic for Open Systems
      1. 6.3.1 Processing the CSV output file to create the base model for the DS8870 storage system
      2. 6.3.2 Migrating the DS8870 storage system to the DS8886 storage system
    4. 6.4 Disk Magic Easy Tier modeling
      1. 6.4.1 Predefined skew levels
      2. 6.4.2 Current workload existing skew level
      3. 6.4.3 Heat map
    5. 6.5 Storage Tier Advisor Tool
      1. 6.5.1 Storage Tier Advisor Tool output samples
      2. 6.5.2 Storage Tier Advisor Tool for Disk Magic skew
  13. Chapter 7. Practical performance management
    1. 7.1 Introduction to practical performance management
    2. 7.2 Performance management tools
      1. 7.2.1 IBM Spectrum Control overview
      2. 7.2.2 IBM Spectrum Control measurement of DS8000 components
      3. 7.2.3 General IBM Spectrum Control measurement considerations
    3. 7.3 IBM Spectrum Control data collection considerations
      1. 7.3.1 Time stamps
      2. 7.3.2 Duration
      3. 7.3.3 Intervals
    4. 7.4 IBM Spectrum Control performance metrics
      1. 7.4.1 DS8000 key performance indicator thresholds
      2. 7.4.2 Alerts and thresholds
    5. 7.5 IBM Spectrum Control reporting options
      1. 7.5.1 WebUI reports
      2. 7.5.2 Cognos reports
      3. 7.5.3 TPCTOOL
      4. 7.5.4 Native SQL Reports
    6. 7.6 Advanced analytics and self-service provisioning
      1. 7.6.1 Advanced analytics
      2. 7.6.2 Self-service provisioning
    7. 7.7 Using IBM Spectrum Control network functions
    8. 7.8 End-to-end analysis of I/O performance problems
    9. 7.9 Performance analysis examples
      1. 7.9.1 Example 1: Disk array bottleneck
      2. 7.9.2 Example 2: Hardware connectivity part 1
      3. 7.9.3 Example 3: Hardware connectivity part 2
      4. 7.9.4 Example 4: Port bottleneck
      5. 7.9.5 Example 5: Server HBA bottleneck
    10. 7.10 IBM Spectrum Control in mixed environments
  14. Part 3 Performance considerations for host systems and databases
  15. Chapter 8. Host attachment
    1. 8.1 DS8000 host attachment
    2. 8.2 Attaching Open Systems hosts
      1. 8.2.1 Fibre Channel
      2. 8.2.2 Storage area network implementations
      3. 8.2.3 Multipathing
    3. 8.3 Attaching z Systems hosts
      1. 8.3.1 FICON
      2. 8.3.2 FICON configuration and sizing considerations
      3. 8.3.3 z/VM, z/VSE, and Linux on z Systems attachment
  16. Chapter 9. Performance considerations for UNIX servers
    1. 9.1 Planning and preparing UNIX servers for performance
      1. 9.1.1 UNIX disk I/O architecture
      2. 9.1.2 Modern approach to the logical volume management in UNIX systems
      3. 9.1.3 Queue depth parameter (qdepth)
    2. 9.2 AIX disk I/O components
      1. 9.2.1 AIX Journaled File System and Journaled File System 2
      2. 9.2.2 Symantec Veritas File System for AIX
      3. 9.2.3 IBM Spectrum Scale
      4. 9.2.4 IBM Logical Volume Manager
      5. 9.2.5 Symantec Veritas Volume Manager
      6. 9.2.6 IBM Subsystem Device Driver for AIX
      7. 9.2.7 Multipath I/O with IBM Subsystem Device Driver Path Control Module
      8. 9.2.8 Symantec Veritas Dynamic MultiPathing for AIX
      9. 9.2.9 Fibre Channel adapters
      10. 9.2.10 Virtual I/O Server
    3. 9.3 AIX performance monitoring tools
      1. 9.3.1 AIX vmstat
      2. 9.3.2 pstat
      3. 9.3.3 AIX iostat
      4. 9.3.4 lvmstat
      5. 9.3.5 topas
      6. 9.3.6 nmon
      7. 9.3.7 fcstat
      8. 9.3.8 filemon
    4. 9.4 Verifying your system
  17. Chapter 10. Performance considerations for Microsoft Windows servers
    1. 10.1 General Windows performance tuning
    2. 10.2 I/O architecture overview
    3. 10.3 Windows Server 2008 I/O Manager enhancements
    4. 10.4 File system
      1. 10.4.1 Windows file system overview
      2. 10.4.2 NTFS guidelines
      3. 10.4.3 Paging file
    5. 10.5 Volume management
      1. 10.5.1 Microsoft Logical Disk Manager
      2. 10.5.2 Veritas Volume Manager
      3. 10.5.3 Determining volume layout
    6. 10.6 Multipathing and the port layer
      1. 10.6.1 SCSIport scalability issues
      2. 10.6.2 Storport scalability features
      3. 10.6.3 Subsystem Device Driver
      4. 10.6.4 Subsystem Device Driver Device Specific Module
      5. 10.6.5 Veritas Dynamic MultiPathing for Windows
    7. 10.7 Host bus adapter settings
    8. 10.8 I/O performance measurement
      1. 10.8.1 Key I/O performance metrics
      2. 10.8.2 Windows Performance console (perfmon)
      3. 10.8.3 Performance log configuration and data export
      4. 10.8.4 Collecting configuration data
      5. 10.8.5 Correlating performance and configuration data
  18. Chapter 11. Performance considerations for VMware
    1. 11.1 Disk I/O architecture overview
    2. 11.2 vStorage APIs for Array Integration support
    3. 11.3 Host type for the DS8000 storage system
    4. 11.4 Multipathing considerations
    5. 11.5 Performance monitoring tools
      1. 11.5.1 Virtual Center performance statistics
      2. 11.5.2 Performance monitoring with esxtop
      3. 11.5.3 Guest-based performance monitoring
      4. 11.5.4 VMware specific tuning for maximum performance
      5. 11.5.5 Workload spreading
      6. 11.5.6 Virtual machines sharing the LUN
      7. 11.5.7 ESXi file system considerations
      8. 11.5.8 Aligning partitions
  19. Chapter 12. Performance considerations for Linux
    1. 12.1 Supported platforms and distributions
    2. 12.2 Linux disk I/O architecture
      1. 12.2.1 I/O subsystem architecture
      2. 12.2.2 Cache and locality of reference
      3. 12.2.3 Block layer
      4. 12.2.4 I/O device driver
    3. 12.3 Specific configuration for storage performance
      1. 12.3.1 Host bus adapter for Linux
      2. 12.3.2 Multipathing in Linux
      3. 12.3.3 Logical Volume Management
      4. 12.3.4 Tuning the disk I/O scheduler
      5. 12.3.5 Using ionice to assign I/O priority
      6. 12.3.6 File systems
    4. 12.4 Linux performance monitoring tools
      1. 12.4.1 Gathering configuration data
      2. 12.4.2 Disk I/O performance indicators
      3. 12.4.3 Identifying disk bottlenecks
  20. Chapter 13. Performance considerations for the IBM i system
    1. 13.1 IBM i storage architecture
      1. 13.1.1 Single-level storage
      2. 13.1.2 Object-based architecture
      3. 13.1.3 Storage management
      4. 13.1.4 Disk pools in the IBM i system
    2. 13.2 Fibre Channel adapters and Multipath
      1. 13.2.1 FC adapters for native connection
      2. 13.2.2 FC adapters in VIOS
      3. 13.2.3 Multipath
    3. 13.3 Performance guidelines for hard disk drives in a DS8800 storage system with the IBM i system
      1. 13.3.1 RAID level
      2. 13.3.2 Number of ranks
      3. 13.3.3 Number and size of LUNs
      4. 13.3.4 DS8800 extent pools for IBM i workloads
      5. 13.3.5 Disk Magic modeling for an IBM i system
    4. 13.4 Preferred practices for implementing IBM i workloads on flash cards in a DS8870 storage system
      1. 13.4.1 Testing environment
      2. 13.4.2 Workloads for testing
      3. 13.4.3 Testing scenarios
      4. 13.4.4 Test results
      5. 13.4.5 Conclusions and recommendations
    5. 13.5 Preferred practices for implementing IBM i workloads on flash cards in a DS8886 storage system
      1. 13.5.1 Testing environment
      2. 13.5.2 Workloads for testing
      3. 13.5.3 Testing scenarios
      4. 13.5.4 Test results
      5. 13.5.5 Conclusions and preferred practices
    6. 13.6 Analyzing performance data
      1. 13.6.1 IBM i performance tools
      2. 13.6.2 DS8800 performance tools
      3. 13.6.3 Periods and intervals of collecting data
    7. 13.7 Easy Tier with the IBM i system
      1. 13.7.1 Hot data in an IBM i workload
      2. 13.7.2 IBM i methods for hot-spot management
      3. 13.7.3 Skew level of an IBM i workload
      4. 13.7.4 Using Easy Tier with the IBM i system
    8. 13.8 I/O Priority Manager with the IBM i system
  21. Chapter 14. Performance considerations for IBM z Systems servers
    1. 14.1 DS8000 performance monitoring with RMF
      1. 14.1.1 RMF Overview
      2. 14.1.2 Direct Access Device Activity report
      3. 14.1.3 I/O response time components
      4. 14.1.4 I/O Queuing Activity report
      5. 14.1.5 FICON host channel report
      6. 14.1.6 FICON Director Activity report
      7. 14.1.7 Cache and NVS report
      8. 14.1.8 Enterprise Disk Systems report
      9. 14.1.9 Alternatives and supplements to RMF
    2. 14.2 DS8000 and z Systems planning and configuration
      1. 14.2.1 Sizing considerations
      2. 14.2.2 Optimizing performance
    3. 14.3 Problem determination and resolution
      1. 14.3.1 Sources of information
      2. 14.3.2 Identifying critical and restrained resources
      3. 14.3.3 Corrective actions
  22. Chapter 15. IBM System Storage SAN Volume Controller attachment
    1. 15.1 IBM System Storage SAN Volume Controller
      1. 15.1.1 SAN Volume Controller concepts
      2. 15.1.2 SAN Volume Controller multipathing
      3. 15.1.3 SAN Volume Controller Advanced Copy Services
    2. 15.2 SAN Volume Controller performance considerations
    3. 15.3 DS8000 performance considerations with SAN Volume Controller
      1. 15.3.1 DS8000 array
      2. 15.3.2 DS8000 rank format
      3. 15.3.3 DS8000 extent pool implications
      4. 15.3.4 DS8000 volume considerations with SAN Volume Controller
      5. 15.3.5 Volume assignment to SAN Volume Controller
    4. 15.4 Performance monitoring
    5. 15.5 Sharing the DS8000 storage system between various server types and the SAN Volume Controller
      1. 15.5.1 Sharing the DS8000 storage system between Open Systems servers and the SAN Volume Controller
      2. 15.5.2 Sharing the DS8000 storage system between z Systems servers and the SAN Volume Controller
    6. 15.6 Configuration guidelines for optimizing performance
    7. 15.7 Where to place flash
    8. 15.8 Where to place Easy Tier
  23. Chapter 16. IBM ProtecTIER data deduplication
    1. 16.1 IBM System StorageTS7600 ProtecTIER data deduplication
    2. 16.2 DS8000 attachment considerations
  24. Chapter 17. Databases for open performance
    1. 17.1 DS8000 with DB2 for Linux, UNIX, and Windows
      1. 17.1.1 DB2 for Linux, UNIX, and Windows storage concepts
    2. 17.2 DB2 for Linux, UNIX, and Windows with DS8000 performance recommendations
      1. 17.2.1 DS8000 volume layout for databases
      2. 17.2.2 Know where your data is
      3. 17.2.3 Balance workload across DS8000 resources
      4. 17.2.4 Use DB2 to stripe across containers
      5. 17.2.5 Selecting DB2 logical sizes
      6. 17.2.6 Selecting the DS8000 logical disk sizes
      7. 17.2.7 Multipathing
    3. 17.3 Oracle with DS8000 performance considerations
      1. 17.3.1 Architecture overview
      2. 17.3.2 DS8000 performance considerations
      3. 17.3.3 Oracle for AIX
    4. 17.4 Database setup with a DS8000 storage system: Preferred practices
      1. 17.4.1 The Oracle stripe and mirror everything approach
      2. 17.4.2 DS8000 RAID policy and striping
      3. 17.4.3 LVM striping
  25. Chapter 18. Database for IBM z/OS performance
    1. 18.1 DB2 in a z/OS environment
      1. 18.1.1 Understanding your database workload
      2. 18.1.2 DB2 overview
      3. 18.1.3 DB2 storage objects
      4. 18.1.4 DB2 data set types
    2. 18.2 DS8000 considerations for DB2
    3. 18.3 DB2 with DS8000 performance recommendations
      1. 18.3.1 Knowing where your data is
      2. 18.3.2 Balancing workload across DS8000 resources
      3. 18.3.3 Solid-state drives
      4. 18.3.4 High-Performance Flash Enclosure
      5. 18.3.5 High Performance FICON and FICON Express 16S
      6. 18.3.6 DB2 Adaptive List Prefetch
      7. 18.3.7 Taking advantage of VSAM data striping
      8. 18.3.8 Large volumes
      9. 18.3.9 Modified Indirect Data Address Words
      10. 18.3.10 Adaptive Multi-stream Prefetching
      11. 18.3.11 DB2 burst write
      12. 18.3.12 DB2 / Easy Tier integration
      13. 18.3.13 Bypass extent serialization in Metro Mirror
      14. 18.3.14 zHyperWrite
      15. 18.3.15 Monitoring the DS8000 performance
    4. 18.4 IMS in a z/OS environment
      1. 18.4.1 IMS overview
      2. 18.4.2 IMS logging
    5. 18.5 DS8000 storage system considerations for IMS
    6. 18.6 IMS with DS8000 performance recommendations
      1. 18.6.1 Balancing workload across DS8000 resources
      2. 18.6.2 Write ahead data set volumes
      3. 18.6.3 Monitoring DS8000 performance
  26. Part 4 Appendixes
  27. Appendix A. Performance management process
    1. Introduction
    2. Purpose
    3. Operational performance subprocess
    4. Tactical performance subprocess
    5. Strategic performance subprocess
  28. Appendix B. Benchmarking
    1. Goals of benchmarking
    2. Requirements for a benchmark
  29. Related publications
    1. IBM Redbooks publications
    2. Other publications
    3. Online resources
    4. How to get IBM Redbooks publications
    5. Help from IBM
  30. Back cover