|
|
United States-English | |
|
|
 |
|
HP-UX System Administrator's Guide: Overview: HP-UX 11i Version 3 > Chapter 2 HP-UX Virtualization Technologies
Load Balancing (One Multiple Core Server, One Operating System Instance) |
|
|
While the HP-UX scheduler does a good job of utilizing multiple cores when they are available, sometimes you need (or want) to override the default scheduling algorithms, for example to dedicate processing resources to an important application. When your server is equipped with more than one core, the technologies in the load balancing category can help you precisely allocate the resources of multiple cores and other processing resources (memory, disk I/O bandwidth). These technologies include: Process Resource Manager (PRM) is a resource management tool used to control the amount of resources that processes use during peak system load (when the server is at 100% core, 100% memory, or 100% disk bandwidth utilization). Processes and users are assigned to PRM Groups. The PRM Groups are then allocated dedicated cores and memory. This both dedicates resources to the groups and isolates the group’s members from using other system resources. The Process Resource Manager can be configured using the HP System Management Homepage or from a series of PRM management and configuration commands. The Process Resource Manager is included in the following operating environments or can be purchased from an HP authorized reseller:
Detailed information about the Process Resource Manager is available on HP’s web site from:
The Process Resource Manager allows you to manually dedicate resources to specific applications. However, many factors are likely to change constantly, such as system load, resource utilization, resource needs, and the mix of applications that are running at any given time. If the conditions and needs of your server environment are constantly changing, WLM can continuously monitor and adjust resources to maintain application performance and business goals. Workload Manager (WLM) is an automatic resource management tool used for goal-based workload management. A workload is a group of processes that are treated as a single unit for the purposes of resource allocation. For example, a database application that consists of multiple cooperating processes could be considered a workload. WLM provides automatic resource allocation and application performance management through the use of prioritized Service Level Objectives (SLOs). Multiple prioritized workloads can be managed dynamically on a single server based on their reported performance levels. WLM manages workloads as defined in a configuration file. You assign applications and users to workload groups. WLM automatically allocates processing core resources to achieve the desired SLO. WLM can manage real memory and disk bandwidth but not in response to SLOs. With real memory, WLM allows you to specify lower and upper limits on the amount of memory a workload receives. You can statically assign disk bandwidth shares. If multiple users or applications within a workload are competing for resources, standard HP-UX resource management determines the resource allocation. The Workload Manager can be configured manually using a configuration file and WLM commands, interactively using the WLM configuration wizard, or by using the WLM graphical user interface. The Workload Manager is available as a standalone product that can be purchased from an HP authorized reseller or is available in the following operating environments:
Detailed information about the Workload Manager is available on HP’s web site at:
Processor Sets, a key technology underlying the Process Resource Manager and the Workload Manager, can also be used on its own. A processor set represents a set of cores grouped together as an independent scheduling domain for use by the HP-UX scheduler for exclusive access to applications assigned to that processor set. Processor Sets allow you to isolate compute intensive or high priority applications from other processes running on a server giving you greater control over the scheduling of critical processes. Processor Sets may be created and re-configured dynamically by users who have the appropriate privileges. The Processor Sets technology can be used alone, but is often used in conjunction with the Process Resource Manager (PRM). See the following documents (available on docs.hp.com) for further information about PSETS:
Instant Capacity technology allows you to purchase at a significantly reduced price: cores, memory, and cell boards that you do not yet own (and cannot yet use). These components, known as iCAP components or components without usage rights, are standing by, ready for activation when you need them. You then purchase usage rights for some or all of these components and are given the ability through the use of a special codeword to activate (temporarily or permanently) these server components. This gives you the ability to quickly handle peak or unexpected demands for server resources as long as the system remains in compliance with the Instant Capacity contract. There are three types of Instant Capacity:
A series of iCAP commands allows you to interact with HP-UX to adjust the processing capacity of a server containing iCAP hardware. The commands allow you to enter purchased codewords to activate your rights to use the stand-by processors, or to decrease the processing resources you are using in order to save money. The commands to administer iCAP resources are:
The capability to activate instant capacity components is part of the HP-UX Foundation Operating Environment (and is therefore available in all operating environments that build upon the Foundation Operating Environment). However, in order to use the instant capacity technologies, you must have purchased and installed components without usage rights to activate. You can find comprehensive information about the various iCAP technologies in the following documents:
Pay per use (PPU) is a product that enables you to “pay as you go”. Pay per use is a pricing model in which you are charged for your actual usage of computing resources. With Pay per use, you treat computer processing capacity as you would electricity, water, or other utilities. You acquire a specific hardware platform with a number of cores and are charged for the actual usage based on one of the following HP contractual agreements:
The Pay per use program uses a dedicated system known as the Utility Meter that is set up by HP. For servers you plan to use with Pay per use, you must set up a software agent (the PPU Agent) on every partition (both hardware and software partitions). The PPU Agent reports usage information to the Utility Meter which in turn communicates to HP your actual usage. A single Utility Meter can service up to 100 servers or partitions running HP-UX 11i or Windows Server 2003. Most of the configuration of the Pay per use components is done through the ppuconfig command. ppuconfig can:
Pay per use is shipped with the HP-UX Foundation Operating Environment and is customer configurable on all HP-UX servers. However, you must enable the Pay per use system to communicate with a Utility Meter (a special system which must be set up by HP). A single Utility Meter can support multiple Pay per use systems (currently up to 100), so you do not need a Utility Meter system for every Pay per use system. You can find comprehensive information about Pay per use and the Utility Meter in the following documents:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 Printable version |
|
|
|
|
|
|||||||||||||||
|
|||||||||||||||
