Getting Started with gWLM

Version A.01.01.x

Manufacturing Part Number: T2762-90005 May 2005 Copyright 2004-2005 Hewlett-Packard Development Company, L.P.

Publication history Second edition May 2005 T2762-90005 HP-UX 11i v1, HP-UX 11i v2 Update 2, and Linux First edition December 2004 T2762-90002 HP-UX 11i v1, HP-UX 11i v2 Update 2, and Linux

Notice
Copyright 2004-2005 Hewlett-Packard Development Company, L.P. All Rights Reserved. Reproduction, adaptation, or translation without prior written permission is prohibited, except as allowed under the copyright laws. The information contained in this document is subject to change without notice. Hewlett-Packard makes no warranty of any kind with regard to this material, including, but not limited to, the implied warranties of merchantability and tness for a particular purpose. Hewlett-Packard shall not be liable for errors contained herein or for incidental or consequential damages in connection with the furnishing, performance or use of this material.

Contents
Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 What is gWLM? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 What are the benets of using gWLM?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 What do I need to know to use gWLM? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 What is the gWLM management model? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 What types of policies are available? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Choosing a policy type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Choosing between an OwnBorrow policy and a utilization policy . . . . . . . . . . . . . . . . 9 Combining the different policy types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Can I see how gWLM will performwithout actually affecting my system? . . . . . . . 10 How do I get started with gWLM? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 How do I stop gWLM? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Show me gWLM in action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 What are some common gWLM tasks? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Fixing the amount of CPU a workload gets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Resizing a workloads npar, vpar, pset, or fss group as needed . . . . . . . . . . . . . . . . . 17 How do I monitor gWLMs effect on my workloads? . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Graphical reports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 High-level view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Other administration tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Initial steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 How do I change from advisory mode to managed mode? . . . . . . . . . . . . . . . . . . . . . 20 How do I create a new policy? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 How do I edit a policy? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 How do I change which policy is associated with a workload? . . . . . . . . . . . . . . . . . . 23 How do I add a new npar / vpar / pset / fss group to an SRD? . . . . . . . . . . . . . . . . . . 24 How do I stop managing a workload? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 How do I stop managing an SRD? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Finding more information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Providing feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Contents

ii

Assumptions

This document presents an overview of the techniques and tools available for using Global Workload Manager, or gWLM. It exposes you to the essentials and allows you to quickly get started using gWLM.

Assumptions
It is assumed that you have already installed the following software: HP Systems Insight Manager HP Global Workload Manager

For information on setting up HP Systems Insight Manager, see the HP Systems Insight Manager Installation and User Guide, available from http://www.hp.com/go/hpsim. If you have not installed gWLM, the following steps give an overview. When you install gWLM, you must: Step 1. Decide which system will be your central management server (CMS), then install the gWLM CMS software on that system. This system must also have HP Systems Insight Manager installed and running. Step 2. Initialize the CMS for gWLM by running gwlminitconfig --initconfig. Step 3. Start the gWLM CMS daemon gwlmcmsd. Step 4. Decide which systems will be your managed nodes, then install the gWLM agent software on those systems. (The agent software is free, but it is functional only for a limited time. For unlimited use, purchase the agent license to use, or LTU, as described in the gWLM Release and Installation Notes.) Step 5. On each managed node, start the gWLM agent daemon gwlmagent. For detailed instructions on installing gWLM for HP-UX and Linux, see the gWLM Release and Installation Notes.

What is gWLM?

What is gWLM?
gWLM is a tool that allows you to centrally dene resource-sharing policies that you can use across multiple HP servers. These policies increase system utilization and facilitate controlled sharing of system resources. gWLMs monitoring abilities provide both real-time and historical monitoring of the resource allocation.

What are the benets of using gWLM?

The benets of gWLM include: Better use of the server capacity you already have Typically, servers are set up with a single workload and ample reserve capacity to handle the peak demand of that workload. gWLM allows you to combine multiple workloads on a single server and make use of the reserve capacity. Condence that your mission-critical workloads get the resources they need Even with multiple workloads on a server, you can ensure your mission-critical workload gets the resources it needs: gWLM automatically adjusts resource allocation, making it easy to share resources when they are plentiful, but also to dedicate those resources to workloads when there are spikes in resource demand. Reduced system administration costs With gWLM managing resource allocation, you can combine more workloads on fewer serversreducing administration costs.

What do I need to know to use gWLM?

What do I need to know to use gWLM?

Here are some terms to know when using gWLM: Workload The collection of processes that are within a single compartment. The compartment can be an nPartition (npar), a virtual partition (vpar), a processor set (pset), or a Fair Share Scheduler (fss) group. gWLM manages a workload by adjusting the system resource allocations for its compartment. (For background information on npars, vpars, psets, and fss groups, see the section What is the gWLM management model? on page 5.) Compartment An npar, a vpar, a pset, or an fss group with its resource allocation being managed by gWLM. Multiple compartmentsall of the same typeare grouped to form a shared resource domain, or SRD. The compartments all share the resources within the SRD. Each compartment holds a workload and can be in only one SRD. gWLM manages each workload by adjusting the resource allocation for its compartment. Shared Resource Domain (SRD) A collection of compartmentsall of the same typethat share system resources. The compartments can be npars, vpars, psets, or fss groups. A server containing npars can be an SRDas long as the requirements in the section What is the gWLM management model? on page 5 are met. A server or an npar divided into vpars can be an SRD for its vpar compartments. Similarly, a server, an npar, or a vpar containing psets can be an SRD for its pset compartments. Lastly, a server, an npar, or a vpar containing fss groups can be an SRD for its fss group compartments. gWLM creates SRDs when you use the Manage New Systems wizard, reached through the menu Optimize -> Global Workload Manager (gWLM) -> Manage New Systems

What do I need to know to use gWLM? Policy A policy is a collection of settings that instruct gWLM how to manage a workloads resources. For example, a policy can indicate the amount of CPU resources a workload owns (and is guaranteed when needed) as well as how much of those resources the workload can lend to other workloads. A single policy can be associated, or applied, to multiple workloads. Mode gWLM offers two modes: advisory and managed. Advisory mode allows you to see what CPU requests gWLM would make for a workloadwithout actually affecting resource allocation. You can use this mode when creating and ne-tuning your policies. Once you are comfortable with your policies, use managed mode to have gWLM automatically adjust the resource allocations for your dened workloads. You can only set the mode on the SRD level: All compartments within an SRD operate in the same mode, either advisory or managed. Deploy Enable gWLM control of an SRD. Deploying an SRD in managed mode enables gWLM control of resource allocation within the SRD. For example, in an SRD based on a vpar that has psets for compartments, deploying an SRD in managed mode allows gWLM to actively migrate CPUs between psets. When deploying an SRD in advisory mode, gWLM simply reports what the allocation would bewithout actually affecting resource allocations on a system. Undeploy Disable gWLMs management of resources in a specied SRD. If an SRD is in managed mode, undeploying stops the migration of system resources between workloads in the SRD. If the SRD is in advisory mode, gWLM no longer provides information on what requests would have been made.

What is the gWLM management model?

What is the gWLM management model?

gWLM was designed to support utility computing across a data center by providing resource-sharing policies that you centrally create and monitor. gWLM moves resources between the workloads in a shared resource domain (SRD) as neededbased on the policies you specify. gWLM allows you to manage resource allocations for several types of system divisions, discussed below. These divisions are referred to as compartments in gWLM. HP-UX Hardware Partitions (npars) A hardware partition, also known as an nPartition or npar, is a physical partition of a server, where each npar runs its own instance of the HP-UX operating system. Using the HP product Instant Capacity, gWLM simulates the movement of CPUs between npars by turning off an active CPU in one npar then turning on a deactivated CPU in another npar in the same complex. Thus, the rst npar has one less active CPU, while the second npar has one additional active CPU. (gWLM maintains the number of active CPUs, honoring the Instant Capacity licensing limits. As a result, no costs are incurred.) HP-UX Virtual Partitions (vpars) A virtual partition is a software partition of a server or of a single nPartition, where each virtual partition runs its own instance of the HP-UX operating system. A virtual partition cannot span an nPartition boundary. Processor sets (psets) A processor set is a collection of processors (CPUs) grouped together for the exclusive access by applications assigned to that processor set. Each application runs only on processors in its assigned processor set. Compartments based on psets are available on HP-UX and Linux. (On Linux, gWLM simulates psets using CPU afnity masks.) HP-UX Fair Share Scheduler groups (fss groups) A group of processes that has its CPU allocation managed by the Fair Share Scheduler that is available with HP-UX. A benet of fss groups is their granularity: You can allocate fractions of CPU resources, rather than only whole CPUs, to the group of processes.

What types of policies are available? For additional information on these partitions, visit: HP Virtual Server Environment web site: http://www.hp.com/go/vse The partitioning continuum and the exible computing environment web site: http://docs.hp.com/hpux/hplex/index.html As noted above, gWLM can manage workloads on both HP-UX and Linux. gWLM manages resources based on the following model: 1. You decide which of your systems you want to manage and what type of compartment you want to use. (For psets and fss groups, gWLM creates the compartments for you.) 2. You assign each workload to a compartment. 3. You associate a policy to the workload indicating how gWLM should allocate resources to the workloads compartment. (gWLM comes with several policies and also lets you dene your own. You can use a single policy for multiple workloads, minimizing the number of policies if desired.) 4. gWLM monitors the CPU consumption of all the workloads in the SRD during the current allocation interval. 5. At the end of the interval, gWLM adjusts the CPU allocations for the compartments in accordance with the policies. 6. gWLM repeats the previous two steps. For information on what types of workloads to combine for optimal resource utilization, see the online help topic Getting the most out of gWLM, available in gWLMs graphical interface in HP Systems Insight Manager.

What types of policies are available?

gWLM allows you to dene several types of policies to instruct gWLM how to manage the resources for your workloads. These types are: Fixed policy This type of policy guarantees that a workloads compartment has a xed (constant) amount of CPU resources.

What types of policies are available? gWLM satises these policies before attempting to satisfy any other type of policies. Utilization policy This type of policy has a target based on utilization. With a CPU utilization policy, gWLM attempts to keep a workloads CPU utilization below the target by adding CPU resources when the workload is using too much of its current CPU allocation. For example, assume a workload has a utilization policy with a target of 80% and an allocation of 5 CPUs. If the workload is consuming 4.5 CPUs, its utilization percentage is 4.5/5, or 90%. gWLM would attempt to allocate additional CPU resources to the workloads compartment to meet the target. An allocation of 6 CPUs would result in a utilization percentage of 4.5/6, or 75%, meeting the target. With a utilization policy, you specify the minimum and maximum CPU requests. Workloads with this type of policy are only guaranteed the minimum request. Utilization policies allow you to prioritize workloads. OwnBorrow policy This type of policy allows you to set the: Amount of CPU resources a workloads compartment owns Minimum amount of CPU resources a workloads compartment must have (after lending resources to other workloads) Maximum amount of CPU resources a workloads compartment can have (after borrowing resources from other workloads)

The compartment of a workload with such a policy is guaranteed the owned amount of CPU when needed. The minimum CPU request and maximum CPU request allow you to specify how much the workload can lend (when resources are not needed) or borrow (when additional resources are needed and available). If a compartment has lent out CPUs and that compartments workload becomes busy, the compartment re-acquires those lent-out CPUs immediately. Custom policy Custom policies are available for advanced users. For information on these policies, see the online help or the gwlmxml(4) man page. You can dene your own policies or use one of the numerous policies that come with gWLM. (You can use one policy for multiple workloads, minimizing the number of policies if desired.)

Choosing a policy type

Choosing a policy type

How do you decide which policy type to use? The table below answers this question for several common use cases. The section following the table helps you decide between using an OwnBorrow policy or a utilization policy. Table 1 Choosing a policy type If... You want gWLM to allocate a constant amount of CPU to a workload IT acts as a service provider to business units Use the following type of policy... Fixed OwnBorrow This policy type allows you to set an owned amount of resources, while also giving you control over how workloads borrow and lend resources. gWLM provides a topborrowers report and a resourceaudit report to help you manage your data center using this model. For more information, see the gwlmreport(1M) man page. You have static vpars but you want to move to a model where CPUs migrate between vpars OwnBorrow For each vpar, set its number of owned CPUs to its static number of CPUs. The vpar is then guaranteed those owned CPUs when needed. OwnBorrow Install the HP product Instant Capacity on each npar. (This software allows gWLM to simulate CPU movement among the npars.) For each npar, set its number of owned CPUs to the number of CPUs you want to guarantee the npar has when needed. You want to tap into a pool of resources taking or giving CPU resources as neededwith no guaranteed access to resources beyond a minimum request Utilization

You have npars but you want to move to a model where CPUs migrate between npars

Choosing a policy type

Choosing between an OwnBorrow policy and a utilization policy

OwnBorrow policies guarantee a certain amount of CPU resources are available to your workload. This amount is based on the owned CPU resources specied in the policy. The workload can lend unneeded resources to other workloadsand get those resources back as soon as they are needed again. With utilization policies, gWLM guarantees CPU resources to the workload based on the minimum CPU value specied in the policy. These resources cannot be shared with other workloads.

Combining the different policy types

Each workload in an SRD must have a policy. You can assign different policy types to the workloads in an SRD as long as the types are compatible. The following table indicates the compatible combinations. Table 2 Policy type Fixed Utilization OwnBorrow Compatible combinations of policy types Fixed Yes Yes Yes Utilization Yes Yes OwnBorrow Yes Yes

Combinations with custom policies are addressed in the online help and the gwlmxml(4) man page.

Can I see how gWLM will performwithout actually affecting my system?

Can I see how gWLM will performwithout actually affecting my system?

gWLM provides an advisory mode that allows you to see how gWLM will approximately respond to a given SRD congurationwithout putting gWLM in charge of your systems resources. Using this mode, you can safely gain a better understanding of how gWLM works. In addition, you can check that your policies behave as expectedwith minimal effect on the system. Once you are comfortable with an SRD, change its mode to managed to let gWLM manage resource allocation for the compartments in the SRD. For information on changing modes, see How do I change from advisory mode to managed mode? on page 20.

How do I get started with gWLM?

gWLM is accessed through HP Systems Insight Manager. After performing the necessary gWLM daemon conguration as described in the gWLM installation notes, the quickest way to start using gWLM to manage new systems is to use the Manage New Systems wizard, as described in the following text. Before you start the wizard though, decide: Which systems you want to manage with gWLM Whether you want to manage your workloads by migrating CPU resources between npars, vpars, processor sets, or fss groups. (CPU migration between npars is simulated using the HP product Instant Capacity, as explained in the section What is the gWLM management model? on page 5.)

10

How do I get started with gWLM? To start the wizard:

NOTE

You must be logged in as root on the systems where you run the mxstart, gwlmcmsd, and gwlmagent commands mentioned below. In HP Systems Insight Manager, you must have authorizations for All Tools or gWLM All Tools.

Step 1. On your gWLM CMS: a. Check that HP Systems Insight Manager is running: # ps -ef | grep mxadmin If there is no match for mxadmin, start HP Systems Insight Manager: # /opt/mx/bin/mxstart b. Start the gWLM CMS daemon if it is not already running: # /opt/gwlm/bin/gwlmcmsd Step 2. On each managed node, start the gWLM agent if it is not already running: # /opt/gwlm/bin/gwlmagent Step 3. Connect to HP Systems Insight Manager by pointing your web browser to: http://hostname:280 where hostname represents the hostname of the CMS. Step 4. Select the menus: Optimize -> Global Workload Manager (gWLM) -> Manage New Systems

11

How do I get started with gWLM? The wizard guides you through the following steps: Step 1. Entering system names You must enter the names of the systems (managed nodes) where the workloads you want gWLM to manage are running. Step 2. Forming the SRDs Based on the systems you entered, gWLM will generate default SRDs and compartments. You can inuence how SRDs are formed by indicating a preference for npar compartments, vpar compartments, pset compartments, or fss group compartments, in case multiple compartments are present on a system.

NOTE

gWLM displays only the compartment types available for the system. For information on making other compartment types available, see the gWLM Release and Installation Notes.

Step 3. Indicating mode of operation for each SRD gWLM allows you to run gWLM in either advisory mode or managed mode. In advisory mode, gWLM only reports how it would migrate CPU resources between compartments. With managed mode, gWLM actually migrates the CPU resources between compartments. You may want to change the SRD names (from what gWLM generates) to be meaningful, thereby simplifying working with the SRDs later. Step 4. Associating, or applying, policies to workloads gWLM comes with a number of policies for: Fixing a workloads CPU allocation to a certain amount of CPU resources Adjusting a workloads CPU allocation based on its CPU utilization Optimizing the borrowing/lending of CPU resources between workloads based on their CPU utilization

You must associate a policy to each workload. You may want to change the workload names (from what gWLM generates) to be meaningful, thereby simplifying working with the workloads later.

12

How do I stop gWLM? Step 5. Review the SRD Verify the SRD is as desired and select the [Finish] button to have gWLM manage the resource allocation for the workloads in the SRD.

How do I stop gWLM?

To turn off gWLM:

NOTE

You must be logged in as root on the systems where you run the gwlmcmsd and gwlmagent commands mentioned below. In HP Systems Insight Manager, you must have authorizations for All Tools or gWLM All Tools.

Step 1. Stop all SRD management, as explained in the section How do I stop managing an SRD? on page 26 Step 2. On the gWLM CMS: a. Stop the gWLM daemon: # /opt/gwlm/bin/gwlmcmsd --stop b. Ensure the le /etc/rc.cong.d/gwlmCtl on HP-UX (/etc/syscong/gwlmCtl on Linux) has GWLM_CMS_START set to 0: GWLM_CMS_START=0 This setting prevents gwlmcmsd from starting automatically when the system boots.

13

Show me gWLM in action Step 3. On each managed node: a. Stop the gWLM daemon: # /opt/gwlm/bin/gwlmagent --stop b. Ensure the le /etc/rc.cong.d/gwlmCtl on HP-UX (/etc/syscong/gwlmCtl on Linux) has GWLM_AGENT_START set to 0: GWLM_AGENT_START=0 This setting prevents gwlmagent from starting automatically when the system boots.

Show me gWLM in action

This section helps you see gWLM move CPUs between vpars. You can use similar steps to see CPUs move between npars, psets, or fss groups. For psets and fss groups, though, you will need to put processes in the desired pset or fss group. (You place processes by using either the Applications tab on the Edit Workloads window or the gwlmplace command.) In this example: The gWLM agent is used on two vpars, which we will call vpar1 and vpar2 These vpars are idle and have a number of unbound CPUs that gWLM can move between them HP Systems Insight Manager and the gWLM CMS software are installed on a vpar called vpar3

14

Show me gWLM in action To see gWLM in action:

NOTE

You must be logged in as root on the systems where you run the mxstart, gwlmcmsd, and gwlmagent commands mentioned below. In HP Systems Insight Manager, you must have authorizations for All Tools or gWLM All Tools.

Step 1. Check that HP Systems Insight Manager is running on the gWLM CMS (vpar3): # ps -ef | grep mxadmin If there is no match for mxadmin, start HP Systems Insight Manager: # /opt/mx/bin/mxstart Step 2. Start the gWLM CMS daemon on vpar3: # vpar3> /opt/gwlm/bin/gwlmcmsd Step 3. Start the gWLM agent daemon on vpar1: # vpar1> /opt/gwlm/bin/gwlmagent Step 4. Start the gWLM agent daemon on vpar2: # vpar2> /opt/gwlm/bin/gwlmagent Step 5. Connect to HP Systems Insight Manager by pointing your web browser to: http://hostname:280 where hostname represents the hostname of the CMS, in this case, vpar3. Step 6. Create a gWLM SRD containing the two vpars by following the steps given in How do I get started with gWLM? on page 10 a. b. c. d. For system names, enter the names of the two vpars separated by a space For the compartment type, select Virtual Partition (vpar) (Optional) Choose a meaningful name to use in place of the generated name for the SRD Ensure the mode is Managed

15

Show me gWLM in action e. Use an OwnBorrow policy for the vpar1 workload and for the vpar2 workload An OwnBorrow policy has a name of the form: Owns_4_CPUs-Max_8_CPUs Ideally, the number of CPUs owned by the workloads will equal the total number of CPUs in the SRD when using this type of policy. (You may need to edit policies to achieve this equality.) f. Conrm and nish the SRD creation

Step 7. View gWLMs real-time reports to show CPU allocation for vpar1 by selecting the following menu in HP Systems Insight Manager: Optimize -> Global Workload Manager (gWLM) -> View Real-time Reports... Step 8. Select the vpar1 workload Step 9. Select the [View Policy Detail] button to see a graph of vpar1s CPU allocation Step 10. Start a CPU-intensive workload in vpar1 If you already have such a workload congured in vpar1, start it now. If you need such a workload, the following command prints its PID (24379 in this case) and then just consumes CPU: # /opt/perl/bin/perl -e print "$$\n";while (1) {}; & [1] 24379 This command will consume most of a single CPU. Start multiple copies of the command to consume additional CPUs. Step 11. Wait a few minutes then look at the Policy Request and Workload Allocation graph from Step 9 to see how the policy for vpar1s workload is requesting CPU allocations and how gWLM is granting them Step 12. Kill the workloads you started in vpar1 Step 13. Repeat Step 7 through Step 12 for the vpar2 workload if you would like to see gWLM move CPUs to vpar2

16

What are some common gWLM tasks?

What are some common gWLM tasks?

gWLM is a powerful tool that allows you to manage your systems in numerous ways. The sections below explain some of the more common tasks that gWLM can do for you.

Fixing the amount of CPU a workload gets

gWLM allows you to give a workload a xed amount of CPU. This xed amount is in the form of a set amount of CPU resources given to an npar, a vpar, a pset, or an fss group. To x the amount of CPU a workload gets, use a xed policy provided by gWLM or create your own. Associate a xed policy with a workload: When creating an SRD, as described in How do I get started with gWLM? on page 10 When adding a workload to an SRD, as described in How do I add a new npar / vpar / pset / fss group to an SRD? on page 24 By changing the policy associated with an existing workload, as described in How do I change which policy is associated with a workload? on page 23

Resizing a workloads npar, vpar, pset, or fss group as needed

To ensure a workload gets the CPU it needswhile also allowing resource sharing when possiblegWLM provides OwnBorrow policies. With such a policy, you indicate how much CPU a workload should own. The workload is then guaranteed this owned amount of CPUwhen it needs it. However, you can congure the workload to: Lend CPU resources to other workloads when it is idle Borrow CPU resources from workloads that are idle

Associate an OwnBorrow policy with a workload: When creating an SRD, as described in How do I get started with gWLM? on page 10 When adding a workload to an SRD, as described in How do I add a new npar / vpar / pset / fss group to an SRD? on page 24 By changing the policy associated with an existing workload, as described in How do I change which policy is associated with a workload? on page 23

gWLMs utilization policies also allow resizing.

17

How do I monitor gWLMs effect on my workloads?

How do I monitor gWLMs effect on my workloads?

For monitoring gWLM, there are several methods, as described below.

Graphical reports
gWLM offers graphs showing either real-time or historical data through HP Systems Insight Manager. For information on interpreting these reports, see the online help. Navigate to real-time reports through the menus: Optimize -> Global Workload Manager (gWLM) -> View Real-time Reports... Navigate to historical reports through the menus: Optimize -> Global Workload Manager (gWLM) -> View Historical Reports

High-level view
For a high-level view of the performance of your SRDs and workloads, navigate through the menus: Optimize -> Global Workload Manager (gWLM) -> View Summaries

18

Other administration tasks

Other administration tasks

This section discusses various administration tasks: How do I change from advisory mode to managed mode? How do I create a new policy? How do I edit a policy? How do I change which policy is associated with a workload? How do I add a new npar / vpar / pset / fss group to an SRD? How do I stop managing a workload? How do I stop managing an SRD?

Initial steps
Several of the tasks below require the same initial steps. These steps are given below.

NOTE

You must be logged in as root on the systems where you run the mxstart, gwlmcmsd, and gwlmagent commands mentioned below.

Step 1. On your gWLM CMS: a. Check that HP Systems Insight Manager is still running: # ps -ef | grep mxadmin If there is no match for mxadmin, start HP Systems Insight Manager: # /opt/mx/bin/mxstart b. Start the gWLM CMS daemon if it is not already running: # /opt/gwlm/bin/gwlmcmsd Step 2. On each managed node, start the gWLM agent if it is not already running: # /opt/gwlm/bin/gwlmagent

19

Other administration tasks

How do I change from advisory mode to managed mode?

Advisory mode allows you to see what CPU requests gWLM would make for a workloadwithout actually affecting resource allocation. Managed mode, however, allows gWLM to automatically adjust the resource allocations for your dened workloads. To change from one mode to the other:

NOTE

In HP Systems Insight Manager, you must have authorizations for All Tools or gWLM All Tools.

NOTE

If you are changing from managed mode to advisory mode and you do not plan to change back soon, be aware that gWLM leaves the npar, vpar, and pset compartments with the CPU counts they had in the last allocation interval. Set the compartments to your desired sizes before changing to advisory mode by associating xed policies with all the compartments and waiting for an allocation interval (15 seconds by default) to pass.

Step 1. Ensure HP Systems Insight Manager, the gWLM CMS daemon (gwlmcmsd) and all the gWLM agents (gwlmagent) are still running, as explained in the section Initial steps on page 19. Step 2. Connect to HP Systems Insight Manager by pointing your web browser to: http://hostname:280 where hostname represents the hostname of the CMS. Step 3. Select, in HP Systems Insight Manger, the menu item Optimize -> Global Workload Manager (gWLM) -> Edit SRDs... Step 4. Select the SRD for which to change the mode Step 5. Select the [Edit] button

20

Other administration tasks Step 6. Scroll down to see the new section that has been added to the page: Edit Shared Resource Domain Step 7. Select the radio button corresponding to the mode you desire Step 8. Select the [OK] button

How do I create a new policy?

A policy instructs gWLM how to manage a workloads resources. To create a policy:

NOTE

In HP Systems Insight Manager, you must have authorizations for All Tools or gWLM All Tools.

Step 1. Ensure HP Systems Insight Manager, the gWLM CMS daemon (gwlmcmsd) and all the gWLM agents (gwlmagent) are still running, as explained in the section Initial steps on page 19. Step 2. Connect to HP Systems Insight Manager by pointing your web browser to: http://hostname:280 where hostname represents the hostname of the CMS. Step 3. Select, in HP Systems Insight Manger, the menu item Optimize -> Global Workload Manager (gWLM) -> Edit Policies... Step 4. Select the [New] button Step 5. Scroll down to see the new section that has been added to the page: New Policy Step 6. Edit the settings, selecting a policy type and specifying the required values and optional values as desired Step 7. Select the [OK] button

21

Other administration tasks

How do I edit a policy?

A policy instructs gWLM how to manage a workloads resources.

NOTE

You can edit the policies provided with gWLM; however, there is currently no way to restore these policies to their original denitions.

To edit a policy:

NOTE

In HP Systems Insight Manager, you must have authorizations for All Tools or gWLM All Tools.

Step 1. Ensure HP Systems Insight Manager, the gWLM CMS daemon (gwlmcmsd) and all the gWLM agents (gwlmagent) are still running, as explained in the section Initial steps on page 19. Step 2. Connect to HP Systems Insight Manager by pointing your web browser to: http://hostname:280 where hostname represents the hostname of the CMS. Step 3. Select, in HP Systems Insight Manger, the menu item Optimize -> Global Workload Manager (gWLM) -> Edit Policies... Step 4. Select the policy to edit Step 5. Select the [Edit] button Step 6. Scroll down to see the new section that has been added to the page: Edit Policy Step 7. Edit the settings Step 8. Select the [OK] button

22

Other administration tasks

NOTE

All workloads associated with this policy will automatically use the updated policy.

How do I change which policy is associated with a workload?

To change the policy affecting how gWLM allocates resources to a workload:

NOTE

In HP Systems Insight Manager, you must have authorizations for All Tools or gWLM All Tools.

Step 1. Ensure HP Systems Insight Manager, the gWLM CMS daemon (gwlmcmsd) and all the gWLM agents (gwlmagent) are still running, as explained in the section Initial steps on page 19. Step 2. Connect to HP Systems Insight Manager by pointing your web browser to: http://hostname:280 where hostname represents the hostname of the CMS. Step 3. Select, in HP Systems Insight Manger, the menu item Optimize -> Global Workload Manager (gWLM) -> Edit Associations... Step 4. Select the workload for which you want to change the policy Step 5. Select the [Change Policy for Workload] button Step 6. Scroll down to see the new section that has been added to the page: Edit Association Step 7. Select the new policy to associate, or apply, to the workload from the pull-down menu Step 8. Select the [OK] button

23

Other administration tasks

How do I add a new npar / vpar / pset / fss group to an SRD?

If you have added an npar or a vpar to your system, set gWLM to manage the new compartment by selecting, in HP Systems Insight Manager, the menu item: Optimize -> Global Workload Manager (gWLM) -> Manage Workloads You can use the same menu item to have gWLM create multiple psets or fss groups for you.

How do I stop managing a workload?

When you stop managing a workload: gWLM stops managing resources for the workload The workloads denition is removed from the SRD, although it remains available for placing in another SRD

NOTE

When gWLM stops managing npar-based or vpar-based workloads, it leaves the npars or vpars with the CPU counts they had in the last allocation interval. For this reason, in Step 4 below, you associate xed policies with the npar-based or vpar-based workloads. For psets and fss groups, gWLM removes the pset or fss group and moves the processes from that compartment to the default compartment.

To stop managing workloads in an SRD: Step 1. Ensure HP Systems Insight Manager, the gWLM CMS daemon (gwlmcmsd) and all the gWLM agents (gwlmagent) are still running, as explained in the section Initial steps on page 19. Step 2. Connect to HP Systems Insight Manager by pointing your web browser to: http://hostname:280 where hostname represents the hostname of the CMS.

24

Other administration tasks Step 3. Associate xed policies with all workloads that you want to unmanage that are based on npars or vpars: a. Select, in HP Systems Insight Manager, the menu item Optimize -> Global Workload Manager (gWLM) -> Edit Associations... b. c. d. e. f. g. Select an npar-based or vpar-based workload to unmanage Select the [Change Policy for Workload] button Scroll down to see the new Edit Association section Select a xed policy Select the [OK] button Repeat Step b through Step f for any additional npar-based or vpar-based workloads to unmanage

Step 4. Wait an allocation interval (15 seconds by default) for gWLM to set CPU allocations based on the xed policies. Step 5. Select, in HP Systems Insight Manger, the menu item Optimize -> Global Workload Manager (gWLM) -> Unmanage Workloads... Step 6. Select one or more workloads a. Select one or more (but not all) workloads you want to unmanage (You cannot select all workloads to unmanage because a deployed SRD must have at least one workload. To stop managing an SRD, undeploy it.) Select the [Next] button

b.

Step 7. Associate policies a. b. Evaluate and change, if needed, the remaining workloads and their associated policies to ensure they are appropriate given that a workload has been removed Select the [Next] button

25

Other administration tasks Step 8. Review summary a. b. c. Review the resulting SRD conguration Select the [Prev] button to make edits Select the [Finish] button if the conguration is as desired and you are ready for gWLM to begin managing the recongured SRD

How do I stop managing an SRD?

To stop gWLM from managing an SRD and its workloads, returning resource allocation to HP-UX:

NOTE

In HP Systems Insight Manager, you must have authorizations for All Tools or gWLM All Tools.

Step 1. Ensure HP Systems Insight Manager, the gWLM CMS daemon (gwlmcmsd) and all the gWLM agents (gwlmagent) are still running, as explained in the section Initial steps on page 19. Step 2. Connect to HP Systems Insight Manager by pointing your web browser to: http://hostname:280 where hostname represents the hostname of the CMS. Step 3. Associate xed policies with all npars or vpars that were in the SRD When gWLM stops managing an SRD, it leaves npar compartments and vpar compartments as they were in the last allocation interval. Associating xed policies allows you to set the sizes exactly to what you want. (psets and fss groups are removed in this situation, with their processes going to the default pset or default fss group.) a. Select, in HP Systems Insight Manager, the menu item Optimize -> Global Workload Manager (gWLM)... -> Edit Associations... b. c. d. 26 Select an npar-based or vpar-based workload to unmanage Select the [Change Policy for Workload] button Scroll down to see the new Edit Association section

Finding more information e. f. g. Select a xed policy Select the [OK] button Repeat Step b through Step f for any additional npar-based or vpar-based workloads to unmanage

Step 4. Select, in HP Systems Insight Manger, the menu item Optimize -> Global Workload Manager (gWLM)... -> Edit SRDs... Step 5. Select the SRD you want to stop managing (undeploy) Step 6. Scroll down to see the new section that has been added to the page: Edit Shared Resource Domain Step 7. Select the radio button corresponding to Undeployed for the SRD State Step 8. Select the [OK] button

Finding more information

The table below indicates the documents that explain various topics. Table 3 Where to nd additional information To... Use gWLM immediately, reading as little as possible See... gWLM Home Page in HP Systems Insight Manager (Optimize -> Global Workload Manager (gWLM) -> Getting Started - gWLM Home) or Concepts topic in online help or Getting started with gWLM (this document) (/opt/gwlm/doc/getting.started.with.gWLM.pdf)

27

Finding more information Table 3 Where to nd additional information (Continued) To... Learn about gWLM concepts See... Concepts topic in online help or Getting started with gWLM (this document) (/opt/gwlm/doc/getting.started.with.gWLM.pdf) Learn gWLM terms What do I need to know to use gWLM? on page 3 or gWLM glossary in online help Learn gWLM best practices Learn about other gWLM features Learn about the gWLM interface in HP Systems Insight Manager Learn about the gWLM command-line interface Learn about gWLM daemons Learn about using secure communications with gWLM Learn more about npars, vpars, and psets Getting the most out of gWLM topic in online help gWLM: Reference and additional topics (/opt/gwlm/doc/gWLM.reference.pdf) online help gwlm(1M) man page gwlmcmsd(1M) man page gwlmsslcong(1M) man page HP Virtual Server Environment web site: http://www.hp.com/go/vse The partitioning continuum and the exible computing environment web site: http://docs.hp.com/hpux/hplex/index.html Learn about using secure communications with gWLM gwlmsslcong(1M) man page

28

Providing feedback

Providing feedback
Email your feedback to the gWLM development team at the following address: [email protected] For a forum with other gWLM users, visit the IT Resource Centers forum for HP-UX Workload/Resource Management: http://forums.itrc.hp.com/cm/CategoryHome/1,,213,00.html

29

Providing feedback

30