Providing Open Architecture High Availability Solutions

Management (CM) services. The operator uses the services of the CM to view the health of each

component, determine which component has failed, and identify the component (field replaceable

unit, FRU) so that it can be replaced. Once the component has been replaced, the CM service is

involved in re-establishing the system model (redundancies, etc.).

Health status may be reported to the manager by the individual components (in the form of event

notifications) or they may be ascertained via a query mechanism from the manager. The manager

may also use testing and diagnostic techniques to determine the health status of an individual

component (or a set of components). These techniques are discussed in Section 6.2 and

Section 6.5.

As system components move from the healthy condition to the degraded condition – where, in the

degraded condition, an abnormality has been detected, but the component is still delivering an

acceptable service level — the system manager can anticipate a system failure in the future.

Several options are available to the system manager at this point, depending on the severity of the

degradation and the policies and procedures, which are in place. Some of these options are:

• Off-loading work from the degraded system component to a more healthy one

• Notifying a system operator

• Performing fault management recovery and repair measures (refer to Section 6.0)

Implicit in this set is the capability of system components to report (via event notifications or in

response to a query) their health status. Also implicit in this set is the capability of a system

manager to receive (via event notifications or queries) this information and process it accordingly.

Monitoring State Changes

Similar to monitoring heath status, it is also important that the CM service monitors changes that

occur in the state of any component. The component states of active/standby as well as the current

operational state need to be recorded. Depending on the application, the CM service may also keep

track of what data object an application is processing, and how far through that processing the

application is. This type of data may also (or alternatively) be stored by an application that is in the

standby mode for the current application.

Testing and Diagnostics

The CM service has the responsibility, though the system model, of understanding which

diagnostic tests can be run on a component. Additionally, it can determine which tests can be run

while the system or component is being used by the system and which must be run only when the

component is not being used. The CM service provides this information to applications and

management middleware which can then run the diagnostics.

In many very high reliability systems, diagnostics are run on components that are not in service to

ensure that they will be ready if needed. Additionally, some systems may run diagnostics

periodically on components, while the are either in use or in standby, to check that they are

working properly.

5.4.5 Techniques

Industry Standard Methods. SNMP, RMON, CIM, TMN(OSI CMIP), IPMI and CORBA are

several of the industry standards available for reporting information external to the system. These

methods may also be used for communicating information between the management function and

the system components. A further discussion of these terms is in Section 5.5.3.