Providing Open Architecture High Availability Solutions

10.1 Collect System Data in Real Time

All critical system components must be continuously monitored and managed in a unified solution.

This includes hardware, software, operating system and applications. Availability management

software, therefore, must include an interface to these components. A flexible approach to use for

this purpose is an object-oriented framework using managed objects.

A managed object is simply a logical representation of a physical or software component. Creating

a managed object includes assigning its name and location, enumerating its relevant attributes, and

defining the methods by which it is to be managed. Based on the defined attributes and methods,

the management middleware can retrieve the appropriate current state information, populate and

dynamically update management database tables, and process the information to determine if any

action is required.

The following components are representative of the types of physical system components that can

be represented by managed objects within management middleware:

Hardware:

• CPU cards

• Line cards

• Fans

• Power supplies

• Temperature/current censors

• Storage devices

• Alarm displays

• Communication links

Software:

• Operating system (and sub-components within)

• Protocol stacks

• Applications

When one of these components is represented by a managed object it is referred to as a managed

component.

The management middleware should also handle the notion of logical system components in

which a group of redundant components act together to provide a single service within the system.

For example, a group of three redundant fans may have a logical managed component to represent

the group.

In addition to collecting data, managed objects provide the ability to control a given component via

methods. Operations include auto-discovery, event handling, and various control functions to

configure or otherwise operate the component. These operations are used by automated service

availability management as well as by external management interfaces such as a web-based user

interface or an external SNMP console.