Jump to content United States-English
HP.com Home Products and Services Support and Drivers Solutions How to Buy
» Contact HP
More options
HP.com home

HP-UX Workload Manager User's Guide: Version A.03.02.02
» 

Technical documentation

Complete book in PDF
» Feedback
Content starts here

 » Table of Contents

 » Glossary

 » Index

spacerspacerspacer
spacer
spacer
   
 

HP Part Number: B8844-90014

Published: January 2007

© Copyright 2000-2007 Hewlett-Packard Development Company, L.P.

Table of Contents

Preface
System platform
Associated software
Document Structure
New in this edition
Feedback to the HP-UX WLM team
Support and patch policies
Training
Notational conventions
Associated documents
1 Introduction
Performance and resource management overview
What is workload management?
What is HP-UX Workload Manager?
Workload management across partitions
Workload management within a single HP-UX instance
Summary
Why use Workload Manager?
Service-level objectives (SLOs)
Prioritized SLOs
WLM and partitions
What is the ideal environment for WLM?
Examples of solutions that WLM provides
SLOs that ensure a specified amount of CPU shares for workloads
SLOs that dynamically allocate resources based on usage goals
Using WLM on multiple servers
Using WLM with HP Integrity Virtual Machines (Integrity VM)
WLM and Process Resource Manager (PRM)
WLM product directory structure
2 WLM quick start: the essentials for using WLM
Network operating environment
WLM shown in action
Where WLM is installed
Seeing how WLM will perform without actually affecting your system
Starting WLM
Stopping WLM
Creating a configuration file
The easiest way to configure WLM
Where to find example WLM configurations
How WLM controls applications
How to put an application under WLM control
Application records: Workload separation by binary name
User records: Workload separation by process owner UID
Unix group records: Workload separation by Unix group ID
Secure compartments: Workload separation by secure compartment
Process maps: Workload separation using your own criteria
prmrun: Starting a process in a workload group
prmmove: Moving an existing process to a workload group
Default: Inheriting workload group of parent process
How to determine a goal for your workload
Some common WLM tasks
Migrating cores across partitions
Providing a fixed amount of CPU resources
Providing CPU resources for a given time period
Providing CPU resources as needed
Other functions WLM provides
Run in passive mode to verify operation
Auditing and billing
Optimize use of Temporary Instant Capacity (TiCAP)
Integrate with various third-party products
Status information WLM provides
Monitoring WLM
ps
wlminfo
wlmgui
prmmonitor
prmlist
GlancePlus
wlm_watch.cgi
Status and message logs
Event Monitoring Service (EMS)
3 How WLM manages workloads
How WLM works
Shares-based SLOs vs goal-based SLOs
How WLM gets application data
How a workload is managed (controllers)
SLO violations
How conflicting SLOs are resolved (arbiter)
Allocating CPU resources: The rising tide model
Example of WLM in use
4 How do I use WLM?
Phasing in your WLM implementation
Steps for using WLM
Reconfiguring WLM
Disabling WLM and its global arbiter
5 Configuring WLM
Configuration file syntactic conventions
Using the WLM configuration wizard
Using the WLM GUI
Tips on using the WLM GUI’s tabs
Controlling system resources
Specifying the WLM parser version
Notification of ‘Instant Capacity needed’ / Pay per use optimization
System-wide settings
Defining the PRM components (optional)
Defining PRM components using the WLM GUI
Specifying workload groups (optional)
Assigning users and user access to workload groups (optional)
Assigning Unix groups to workload groups (optional)
Assigning applications to workload groups (optional)
Assigning secure compartments to workload groups (optional)
Specifying process maps to define your own criteria for workload separation (optional)
Specifying disk bandwidth shares (optional)
Specifying a group’s minimum CPU resources (optional)
Specifying a group’s maximum CPU resources (optional)
Weighting a group so it gets more CPU resources (optional)
Specifying a group’s minimum memory (optional)
Specifying a group’s maximum memory (optional)
Weighting a group so it gets more memory (optional)
Defining SLOs
Defining SLOs using the WLM GUI
Specifying the SLO name (required)
Specifying the priority (required)
Specifying the workload group to which the SLO applies (optional)
Specifying the lower and upper bound requests on CPU resources (optional)
Specifying a goal (optional)
Specifying a fixed or additive allocation of CPU shares (optional)
Specifying a shares-per-metric allocation request (optional)
Specifying when the SLO is active (optional)
Tuning the metrics and the SLOs
Tuning WLM using the WLM GUI
Specifying a data collector (optional)
Specifying the WLM interval (optional)
Using absolute CPU units
Distributing excess CPU resources to your workloads (optional)
Refining granularity of CPU (and memory) allocation by increasing shares per core (optional)
Temporarily removing groups with inactive SLOs (optional)
Capturing your collectors’ stderr (optional)
Smoothing metric values (optional)
Controlling averaging in usage controllers (optional)
Trimming the statistics log file automatically (optional)
Tuning a workload’s SLO convergence:
cntl_kp (optional)
Tuning a workload’s SLO convergence: cntl_convergence_rate (optional)
Tuning the goal buffer (optional)
Releasing cores properly (optional)
Example configuration
Trying a configuration without affecting the system
Passive mode versus actual WLM management
Activating the configuration file
Setting WLM to start automatically at reboot
Setting WLM global arbitration to start automatically at reboot
Setting the WLM communications daemon to start automatically at reboot
Securing WLM communications
Enabling statistics logging at reboot
Disabling statistics logging
WLM and kernel parameters
6 Auditing and billing
Example wlmaudit report
Audit data files
Enabling auditing at reboot
7 Managing SLOs across partitions
Overview
Recommendations, requirements, and restrictions
Setting up cross-partition management
Setting up your WLM configuration file
Setting up your WLM global arbiter configuration file
par structure
Using the WLM GUI to set up the global arbiter configuration file
Specifying the global arbiter interval (optional)
Specifying the port (optional)
Trimming the global arbiter statistics log automatically (optional)
Specifying the priority at which to use Temporary Instant Capacity or Pay per use resources (optional)
Specifying the reserve threshold that determines when WLM stops activating temporary capacity resources
8 Management of nested nPars / vPars / workload groups
Setting up the various configurations
Managing FSS and PSET-based groups inside vPars inside nPars (Instant Capacity available in the complex)
Managing FSS and PSET-based groups inside vPars inside nPars (Instant Capacity not available)
Managing FSS and PSET-based groups inside vPars
9 Example configuration files
distribute_excess.wlm
enabling_event.wlm
entitlement_per_process.wlm
fixed_entitlement.wlm
manual_cpucount.wlm
manual_entitlement.wlm
metric_condition.wlm
par_manual_allocation.wlm
par_manual_allocation.wlmpar
par_usage_goal.wlm
par_usage_goal.wlmpar
performance_goal.template
stretch_goal.wlm
time_activated.wlm
transient_groups.wlm
twice_weekly_boost.wlm
usage_goal.wlm
usage_stretch_goal.wlm
user_application_records.wlm
10 Monitoring SLO compliance and WLM
Monitoring WLM with the wlminfo command
Monitoring WLM with the wlmgui command
Monitoring the configuration
Monitoring the number of CPU resources
Monitoring the workloads
Monitoring SLOs
Monitoring items you define
Monitoring WLM with EMS
What EMS resources are available?
Configuring EMS notification
A WLM command reference
wlmaudit
wlmcert
wlmcomd
wlmcw
wlmd
wlmgui
wlminfo
wlmpard
wlmrcvdc
wlmsend
B WLM configuration file
syntax overview
Configuration file syntax
Configuration file example
C HP-UX command and system call support
D Integration with other products
Integrating with Process Resource Manager (PRM)
Integrating with processor sets (PSETs)
Integrating with nPartitions (nPars)
Integrating with virtual partitions
Integrating with HP Integrity Virtual Machines (Integrity VM)
Running WLM on an Integrity VM Host
Running WLM within an Integrity VM (guest)
For more HP Integrity VM information
Integrating with Temporary Instant Capacity (TiCAP)/ Pay per use (PPU)
For more Temporary Instant Capacity / Pay per use information
Integrating with Security Containment (to form Secure Resource Partitions)
Integrating with
OpenView Performance Agent (OVPA) / OpenView Performance Manager (OVPM)
Integrating with Serviceguard
Steps for integration
Integrating with HP Systems Insight Manager (SIM) and Servicecontrol Manager (SCM)
What WLM tasks are available through SIM / SCM?
Accessing the WLM tools
For more SCM information
Integrating with Oracle® databases
Why use Oracle database metrics with WLM?
Tools in the HP-UX WLM Oracle Database Toolkit (ODBTK)
What metrics are available?
How do I get started with ODBTK?
ODBTK examples
For more ODBTK information
Integrating with Apache
Why use ApacheTK?
Tools in ApacheTK
How do I get started with ApacheTK
For more ApacheTK information
Integrating with BEA WebLogic Server
Why use WebLogicTK?
Tools in WebLogicTK
How do I get started with WebLogicTK
For more WebLogicTK information
Integrating with SAP software
Why use SAPTK?
Tools in SAPTK
How do I get started with SAPTK?
For more SAPTK information
Integrating with SAS software
Why use SASTK?
Tools in SASTK
How do I get started with SASTK?
For more SASTK information
Integrating with the HP-UX SNMP agent
Why use SNMPTK?
Tools in SNMPTK
How do I get started with SNMPTK?
For more SNMPTK information
E Useful PRM utilities
F Understanding how PRM manages resources
Management of CPU resources
Management of CPU resources for real-time processes
Management of real memory
Capping memory use
Management of locked memory
Management of disk bandwidth
How resource allocations interact
Management of applications
How application processes are assigned to workload groups at start-up
How child processes are handled
Pattern matching for file names
Pattern matching for renamed application processes
How the application manager affects workload group assignments
G Migrating from PRM to WLM
H Advanced WLM usage: Using performance metrics
Configuring WLM for metric-based SLOs
Overview
Specifying a metric goal (optional)
Specifying a shares-per-metric allocation request (optional)
Specifying a data collector
Capturing your collectors’ stderr (optional)
Smoothing metric values (optional)
Supplying data to WLM
How applications can make metrics available to WLM
What happens when there is no new data?
I have this type of data—how do I send it to WLM?
What methods exist for sending data to WLM?
Handling signals in data collectors
Glossary
Index

List of Figures

3-1 WLM overview
3-2 CPU allocation: the rising tide model
3-3 Server without WLM
3-4 Server with WLM
5-1 HP-UX WLM Configuration Wizard
5-2 HP-UX WLM GUI
5-3 Workloads with SLOs at same priority
5-4 Additional workload with same priority SLO
5-5 Usage goal conceptually
5-6  Effect of the cntl_margin tunable parameter
5-7 Normal WLM operation
5-8 Passive WLM operation
7-1 WLM managing partitions
8-1 Nested partitions to be managed (with Instant Capacity)
8-2 Nested partitions to be managed (without Instant Capacity)
10-1 Monitoring a configuration
10-2 Monitoring the number of CPU resources
10-3 Monitoring a workload
10-4 Items to graph when monitoring a workload
10-5 Graphing SLOs
10-6 Custom graphing
10-7 Interaction between WLM and EMS
D-1 WLM and Virtual Machines
D-2 WLM integration with Serviceguard
F-1 Locked memory distribution by prm2d memory manager
H-1 Interaction of WLM and ARM-instrumented applications
H-2 wlmrcvdc: command-pipe mode (wlmrcvdc command [args])
H-3 wlmrcvdc: FIFO-file mode (wlmrcvdc)
H-4 wlmsend: command-output mode (command | wlmsend metric)
H-5 wlmsend: single-value mode (wlmsend metric value)

List of Tables

1-1 Performance and resource utilization monitoring methods
1-2 Performance controlling methods
1-3 WLM directories and files
2-1 WLM installation directories
2-2 Example WLM configurations
5-1 weight keyword example (weights take effect)
5-2 weight example with distribute_excess = 0 (off)
5-3 weight example with distribute_excess = 1 (on)
5-4 memweight example
5-5 Capturing remaining CPU resources with stretch goals
10-1 Overview of WLM EMS resources
C-1 HP-UX commands/system calls that support workload groups
C-2 WLM options in HP-UX commands
D-1 ODBTK’s example WLM configuration files described
D-2 wlmoradc example configuration files described
F-1 Resources managed
F-2 Group assignments at process start-up
H-1 Type of data and its transport method
H-2 Available transport methods
H-3 ARM API routines
   


Preface  
Printable version
Privacy statement Using this site means you accept its terms Feedback to webmaster
© 2000-2007 Hewlett-Packard Development Company, L.P.