Why use Workload Manager?

The traditional open systems usage model has been one application running per server, which leads to surplus capacity per server and a proliferation of servers with too many servers to manage. Typically, each server is sized to provide headroom for peak capacity and future growth. As servers are introduced into this traditional environment, surplus capacity grows without providing any opportunity to share this excess capacity among the applications.

WLM, in collaboration with other HP Virtual Server Environment tools, significantly lessens the need to provide the same degree of surplus capacity. WLM allows the surplus capacity to be shared among multiple applications on the same system. You can configure WLM to ensure that each application runs at the performance level required to meet your business goals. By enabling you to consolidate your data centers and multiple applications onto fewer servers, WLM significantly reduces your administration and computer resource costs.

WLM provides you the ability to:

Run multiple workloads on a single system and maintain performance of each workload
Prioritize workloads on a single system, adjusting the CPU allocations based on each workload’s goals
Ensure that critical workloads have sufficient resources to perform at desired levels
Manage by service-level objectives (SLOs) within and across virtual partitions or nPartitions
Adjust resource allocations by automatically enabling or disabling SLOs based on time of day, system events, or application metrics
Enable SLOs associated with an HP Serviceguard package failover
Adjust the number of cores in a partition or a PSET to meet SLOs
Grant a workload dedicated CPU and memory resources in the form of a PSET
Assign Secure Resource Partitions (created by the HP-UX Security Containment feature) to workloads based on PSETs or FSS groups, giving the workloads isolation and automatic resource allocation
Grant a workload CPU resources in direct proportion to a metric, such as number of processes in the workload
Set minimum and maximum amounts of CPU resources available to a workload’s applications
Set minimum and maximum amounts of memory available to a workload’s applications
Set and manage user expectations for performance
Monitor resource consumption by applications or users through HP GlancePlus, WLM utilities, or PRM utilities

When configuring WLM, you define workloads, associate applications with the workloads, and then delineate SLOs for each workload. The SLOs specify a workload’s necessary usage and performance goals. WLM helps the workloads meet the desired goals by dynamically allocating resources based on workload performance, varying workloads, and changing system conditions. Thus, WLM ensures that performance levels are acceptable and business goals are achieved. Information about configuring WLM is provided in Chapter 5 “Configuring WLM”. Information about configuring WLM to manage resources and application performance across partitions is provided in Chapter 7 “Managing SLOs across partitions”.

Service-level objectives (SLOs)

A key reason for using WLM is its ability to manage service-level objectives. After defining a workload, you can specify one or more SLOs for each workload. WLM allocates CPU resources to workloads based on whether the application in the workload is underperforming, meeting, or overperforming its SLOs.

SLOs can be shares-based or goal-based. With a shares-based SLO, WLM tries to grant the associated workload a fixed amount of the CPU resources by allocating CPU shares for the workload. (Each CPU share is 1/100 of a single core or 1/100 of total cores on a system, depending on the WLM mode of operation, as explained in “Using absolute CPU units”.)

With a goal-based SLO, WLM actively changes the CPU allocation of the associated workload to best meet the SLO. Goal-based SLOs are based on one of two types of goals:

Usage Goals
Goals based on a workload’s utilization of its allocated CPU resources. If the processes in a workload are not using a certain amount of the workload’s allocation, the allocation is decreased; if the processes are using a high percentage of the workload’s allocation, the allocation is increased.
Metric goals
Goals based on a metric, such as processing at least x transactions per minute or a response time under y seconds. Metric goals are based on performance data and require understanding of that data. HP recommends using usage goals, as usage goals can be implemented immediately without prior knowledge of workload performance.

A goal-based SLO consists of:

A workload
A goal
A priority ranking
Optional conditions, such as time of day or a particular event
Optional lower and/or upper bounds for CPU resources

A shares-based SLO consists of the same elements except it does not include a goal but rather a shares allocation.

For more information comparing shares-based and goal-based SLOs, see “Shares-based SLOs vs goal-based SLOs”.

Prioritized SLOs

Another important reason for using WLM is that it allows you to prioritize the SLOs. When CPU resources are not sufficient to satisfy all SLOs, WLM grants CPU resources to the highest priority SLOs first. After the demands of the higher priority SLOs are satisfied, WLM grants the lower priority SLOs any resources that remain available. Valid priorities start at 1, with 1 being the highest priority.

SLO priorities do not have to be uniquely assigned—multiple SLOs can be granted the same priority, allowing more than one workload’s objective to be top priority. This can be beneficial when multiple workloads are equally important. Typically, all the SLOs in a given configuration should not be assigned the same priority; otherwise, under a heavy system load, WLM may not be able to allocate CPU resources effectively when there is not enough resources to satisfy all SLOs.

A single workload can have multiple SLOs, each with a different priority. One SLO would be the high priority, “must meet” goal and the remaining SLOs would be lower priority, “meet if possible” goals. For example, a priority 1 goal might be associated with a virtual partition to maintain an allocation of at least two cores. A priority 2 goal might be to allocate three or four cores if available when the workload becomes very busy. This lower priority goal (a “stretch goal”) is met only after the priority 1 SLOs for this workload and all other workloads are met.

For information on how WLM uses priorities, see “Allocating CPU resources: The rising tide model”.

Why use Workload Manager?

Technical documentation

» Table of Contents

» Glossary

» Index

Service-level objectives (SLOs)

Prioritized SLOs