Resource Optimized High Availability in Cloud
Resource Optimized High Availability (ROHA) in Cloud is a function in PowerHA® SystemMirror® that automatically and dynamically manages Dynamic Logical Partitions (DLPARs) resources. You can configure ROHA in Cloud with virtual server instances in Cloud, hardware resource provisioning, and cluster tunable configurations.
PowerHA SystemMirror 7.2.6 SP1 supports ROHA feature with a standard cluster. Starting with PowerHA SystemMirror 7.2.7, it also supported site-based cluster. The Shared processor pool (SPP) feature is supported with ROHA in Cloud.
Starting with the PowerHA SystemMirror 7.2.8 and later, the cloud administrator must create a shared processor pool and configure the PowerVS instances by selecting the shared processor pool.
PowerHA SystemMirror7.2.8 does not require special configuration. A cloud administrator must select SPP when configuring a virtual machine. The IBM infrastructure team controls SPP when allocating the resources to the virtual machine.
For more information about shared processor pool, see IBM Cloud documentation.
For a reference example of shared processor pool, see IBM Cloud documentation.
Before you use ROHA in Cloud, you must add all virtual server instances that are in Cloud onto all associated clusters in PowerHA SystemMirror. You must also plan for the necessary resources for your applications, identify your workloads and your requirements for physical resources (CPU cores, virtual CPUs, and memory). After you identify all these requirements, you must configure ROHA in Cloud.
- Install the following applications on your cluster.
- Python Interpreter version 2.7.x
- Representational State Transfer (REST) application programming interface (API) for AIX® module (libcurl.a) on Cloud logical partitions (LPARs)
- Add Cloud instance details for all the associated cluster nodes by using the clmgr manage cluster cloudroha command. PowerHA SystemMirror automatically collects and adds Cloud instance details for all the associated cluster nodes.
- Create hardware resource provisioning for each application controller that you identified for your workloads by running the clmgr add roha command for cluster.
- Define virtual server instances in Cloud at the site level or cluster level if your topology requires them.
- Change the default value of the virtual server tunable such as the timeout value for DLPAR operations.
- Verify virtual server instances in Cloud that are used by the cluster and by each node.
- Change the clusters ROHA tunables for DLPAR acquisition and release operations.