HACMP integration with the CUoD function
By integrating with DLPAR and CUoD, HACMP ensures that each node can support the application with reasonable performance at a minimum cost. This way, you can upgrade the capacity of the logical partition in cases when your application requires more resources, without having to pay for idle capacity until you actually need it.You can configure cluster resources so that the logical partition with minimally allocated resources serves as a standby node, and the application resides on another LPAR node that has more resources than the standby node. This way, you do not use any additional resources that the frames have until the resources are required by the application.When it is necessary to run the application on the standby node, HACMP ensures that the node has sufficient resources to successfully run the application. The resources can be allocated from two sources:- The free pool. The DLPAR function provides the resources to the standby node, by allocating the resources available in the free pool on the frame.
- CUoD provisioned resources. If there are not enough available resources in the free pool that can be allocated through DLPAR, the CUoD function provides additional resources to the standby node, should the application require more memory or CPU.
A typical HACMP cluster to use with the CUoD functionYou can configure an HACMP cluster within one or more System p servers, using two or more logical partitions. You can also configure a cluster on a subset of LPARs within one frame. Or, the cluster can use partitions from two or more frames, where the nodes can be defined as a subset of LPARs from one frame and a subset of LPARs from another frame, all connected to one or more HMCs. The following figure illustrates a typical two-frame configuration:
Terminology for resource types and memory allocation
The following terms are used in this guide. They help you to distinguish between different types of resources allocation that can occur in an HACMP cluster that uses DLPAR and CUoD functions.- Total amount of resources, or permanent resources. The number of CPUs and the amount of memory that are physically available for use by all LPARs on the frame. This amount includes all permanent, or paid for, resources and may also include those CUoD resources that have already been paid for.
- Free pool. The number of CPUs and the amount of memory that can be dynamically allocated by HACMP through HMC to the LPAR, should the LPAR require additional resources. The free pool is the difference of the total amount of resources on the frame minus the resources that are currently used by the LPARs.Note: The free pool includes resources on a particular frame only. For instance, if a cluster is configured with LPARs that reside on frames A and B, HACMP does not request resources from a pool on frame B for an LPAR that resides on frame A.
- CUoD pool. The number of CPUs and the amount of memory that can be allocated by HACMP using the CUoD license, should the LPAR require additional resources. The CUoD pool depends on the type of CUoD license you have.Note: The CUoD pool includes resources on a particular frame only.
- LPAR minimum amount. The minimum amount (or quantity) of a resource, such as CPU or memory, that an LPAR requires to be brought online or started. The LPAR does not start unless it meets the specified LPAR minimum. When DLPAR operations are performed between LPARs, the amount of resources removed from an LPAR cannot go below this value. This value is set on the HMC and is not modified by HACMP. Use the lshwres command on the HMC to verify this value.
- LPAR desired amount. The desired amount of a resource that an LPAR acquires when it starts, if the resources are available. This value is set on the HMC and is not modified by HACMP. Use the lshwres command on the HMC to verify this value.
- LPAR maximum amount. The maximum amount (or quantity) of a resource that an LPAR can acquire. When DLPAR operations are performed, the amount of resources added to an LPAR cannot go above this value. This value is set on the HMC and is not modified by HACMP. Use the lshwres command on the HMC to verify this value.
The following figure illustrates the relationship between the total amount of memory and resources on the frame (server), the free pool, and the resources that could be obtained through CUoD:
By integrating with DLPAR and CUoD, HACMP ensures that each node can support the application with reasonable performance at a minimum cost. This way, you can upgrade the capacity of the logical partition in cases when your application requires more resources, without having to pay for idle capacity until you actually need it.
You can configure cluster resources so that the logical partition with minimally allocated resources serves as a standby node, and the application resides on another LPAR node that has more resources than the standby node. This way, you do not use any additional resources that the frames have until the resources are required by the application.
When it is necessary to run the application on the standby node, HACMP ensures that the node has sufficient resources to successfully run the application. The resources can be allocated from two sources:
- The free pool. The DLPAR function provides the resources to the standby node, by allocating the resources available in the free pool on the frame.
- CUoD provisioned resources. If there are not enough available resources in the free pool that can be allocated through DLPAR, the CUoD function provides additional resources to the standby node, should the application require more memory or CPU.
A typical HACMP cluster to use with the CUoD function
You can configure an HACMP cluster within one or more System p servers, using two or more logical partitions. You can also configure a cluster on a subset of LPARs within one frame. Or, the cluster can use partitions from two or more frames, where the nodes can be defined as a subset of LPARs from one frame and a subset of LPARs from another frame, all connected to one or more HMCs. The following figure illustrates a typical two-frame configuration:
Terminology for resource types and memory allocation
The following terms are used in this guide. They help you to distinguish between different types of resources allocation that can occur in an HACMP cluster that uses DLPAR and CUoD functions.
- Total amount of resources, or permanent resources. The number of CPUs and the amount of memory that are physically available for use by all LPARs on the frame. This amount includes all permanent, or paid for, resources and may also include those CUoD resources that have already been paid for.
- Free pool. The number of CPUs and the amount of memory that can be dynamically allocated by HACMP through HMC to the LPAR, should the LPAR require additional resources. The free pool is the difference of the total amount of resources on the frame minus the resources that are currently used by the LPARs.Note: The free pool includes resources on a particular frame only. For instance, if a cluster is configured with LPARs that reside on frames A and B, HACMP does not request resources from a pool on frame B for an LPAR that resides on frame A.
- CUoD pool. The number of CPUs and the amount of memory that can be allocated by HACMP using the CUoD license, should the LPAR require additional resources. The CUoD pool depends on the type of CUoD license you have.Note: The CUoD pool includes resources on a particular frame only.
- LPAR minimum amount. The minimum amount (or quantity) of a resource, such as CPU or memory, that an LPAR requires to be brought online or started. The LPAR does not start unless it meets the specified LPAR minimum. When DLPAR operations are performed between LPARs, the amount of resources removed from an LPAR cannot go below this value. This value is set on the HMC and is not modified by HACMP. Use the lshwres command on the HMC to verify this value.
- LPAR desired amount. The desired amount of a resource that an LPAR acquires when it starts, if the resources are available. This value is set on the HMC and is not modified by HACMP. Use the lshwres command on the HMC to verify this value.
- LPAR maximum amount. The maximum amount (or quantity) of a resource that an LPAR can acquire. When DLPAR operations are performed, the amount of resources added to an LPAR cannot go above this value. This value is set on the HMC and is not modified by HACMP. Use the lshwres command on the HMC to verify this value.
The following figure illustrates the relationship between the total amount of memory and resources on the frame (server), the free pool, and the resources that could be obtained through CUoD:
Tidak ada komentar:
Posting Komentar