Managing a policy-managed database requires less configuration and reconfiguration steps than an administrator-managed one with respect to creation, sizing, patching, and load balancing. Also, because any server in the server pools within the cluster can run any of the databases, you do not have to create and maintain database instance-to-node-name mappings. You can perform maintenance tasks such as patching by relocating servers into the Free pool or by adjusting the server pool minimum and maximum sizes, thereby retaining required availability.
Policy-managed databases also facilitate the management of services, because they are assigned to a single server pool and run as singletons or uniform across all servers in the pool. You no longer have to create or maintain explicit preferred and available database instance lists for each service. If a server moves into a server pool because of manual relocation or a high availability event, all uniform services and their dependent database instances are automatically started. If a server hosting one or more singleton services goes down, those services will automatically be started on one or more of the remaining servers in the server pool. In the case of Oracle RAC One Node, the corresponding database instance will also be started automatically.
Managing services relative to each other is improved by making use of the importance attribute of each server pool. Each service running in a server pool inherits the server pool's importance relative to the other server pool-hosted services in the cluster. If the minimum size of the most important server pool is greater than zero, then the services and associated database instances in that server pool are started first on cluster startup and will be the last services and database instances running, as long as there is one server running in the cluster. You can offer services not critical to the business in the least important server pool, ensuring that, should sufficient resources not be available due to demand or failures, those services will eventually be shut down and the more business-critical services remain available.
Because many management tasks may involve making changes that can affect multiple databases, services, or server pools in a consolidated environment, you can use the evaluate mode for certain SRVCTL commands to get a report of the resource impact of a command.
Consider the following example, that evaluates the effect on the system of modifying a server pool:
$ srvctl modify srvpool -l 3 -g online -eval Service erp1 will be stopped on node test3 Service reports will be stopped on node test3 Service inventory will be stopped on node test3 Service shipping will be stopped on node test3 Database dbsales will be started on node test3 Service orderentry will be started on node test3 Service billing will be started on node test3 Service browse will be started on node test3 Service search will be started on node test3 Service salescart will be started on node test3 Server test3 will be moved from pool backoffice to pool online
As shown in the preceding example, modifying a server pool can result in many resource state changes. You can use a policy set through either Oracle Clusterware or Oracle Database Quality of Service Management.
"SRVCTL Usage Information" for a list of SRVCTL commands which use the evaluate mode
Oracle Clusterware Administration and Deployment Guide for information about the Oracle Clusterware policy set
Oracle Database Quality of Service Management User's Guide for information about the Oracle Database Quality of Service Management policy set