Optimize Distributed Workload - Expert Oracle RAC Performance Diagnostics and Tuning - page 488

Distributed Workload Management (DWM)
Workloads on systems can be different depending on the day of the month or they can also vary based on the period
of the year. To ensure sufficient resources are available based on demand for resources, they need to be balanced
through some kind of internal mechanism, for example, prioritizing resources to the order processing systems during
holiday season when customers tend to place online orders and at the same time, not limiting the resources required
for the non-seasonal applications. Similarly, the payroll application would require more resources during payroll
processing, which could be either weekly, biweekly or monthly. All these varying seasonal demands for resources
could be a challenge for several organizations.
Oracle has enhanced and integrated some of its existing technology into RAC architecture for efficient workload
management by allowing the control and allocation of resources required by the various processes and by efficiently
managing them and allocating them based on the priority and importance. Two main areas of incorporation
are included.

Resource Manager
Provisioning resources to database users, applications, or services in an Oracle database allows DBAs to control
the allocation of available resources between the various users, applications, or services. This ensures each user,
application, or service gets a fair share of the available computing resources. This is achieved by determining
predefined resource plans that allocate resources to various consumer groups based on resource usage criteria such
as CPU utilization or number of active sessions.

Fast Application Notification (FAN)
Traditionally, applications connect to the database based on user requests to perform an operation such as retrieve or
update information. During the process of connecting to the database, if the database is not accepting any connection
requests because a node, instance, database, service, or listener is down, the connection manager will have to
return an error back to the application, which in turn will determine the next action to be taken. In the case of a RAC
implementation, the next step (if the node, instance, or listener is down) would be to attempt connecting to the next
available instance on the host address, defined in the address list in the TNS connection descriptor. The time taken to
react to such connection failures and the application to subsequently retry the connection to another node, instance,
or listener is often long.
FAN was introduced in Oracle Database 10g RAC to proactively notify applications regarding the status of the
cluster and any configuration changes that take place. FAN uses the Oracle notification services (ONS) for the actual
notification of the event to its other ONS clients. As illustrated in Figure 15-2 , ONS provides and supports several
callable interfaces that could be used by different applications to take advantage of the HA solutions offered in
Oracle RAC. Applications supported and illustrated in Figure 15-2 include:

• Java Database Connectivity (JDBC)

• Oracle Notification Service (ONS) Application Programming Interface(API)

• Oracle Call Interface (OCI)

• Oracle Data Provider(ODP).NET Oracle Data Provider

• Simple Network Management Protocol(SNMP) Console

• Simple Mail Transfer Protocol (SMTP) email services