IBM Books

Administration Guide


Understanding network management

The Simple Network Management Protocol (SNMP) describes how management data is packaged and transported through the network in order for SNMP agents and SNMP network managers to communicate. The Management Information Base (MIB) defines the structure of the data being transported.

Understanding SNMP

SNMP uses a client-server approach to management, defining two roles: Manager (client) and Agent (server).

The Manager (NetView for AIX is an example) oversees the overall network activity. Critical events in the form of SNMP traps are sent to this Manager to alert network operators of problems in the network. The Manager also acts as the central point for storing and displaying statistics maintained by the Agent (server). These statistics are maintained by the Agent in the form of MIB variables.

The Agent is responsible for reporting on and maintaining the data pertaining to a managed resource, such as a device or, in the case of the SP Agent, a set of configuration information identifying the nodes comprising the SP system and tables of resource attributes within the SP which can be monitored. Agents can run on several different types of managed nodes.

Understanding the MIB

Every Agent supports a Management Information Base (MIB), a set of variables that represent the physical and logical resources of the managed systems or Agent. The MIB is not a database, in the sense of a monolithic collection of data, but rather it represents dynamic information. The values of the variables are maintained by different system functions such as the kernel, device drivers, or subsystems. The Agent obtains the variable values from these system functions.

The Manager can read variable values via these SNMP requests:

GET
Requests the SNMP Agent to retrieve the value of the specified variable and return it to the requester (the Manager).

GET NEXT
Requests the SNMP Agent to retrieve the value of the next variable, after the one specified in the request, and return it to the Manager. This is especially useful for retrieving tabular information, and multiple variables can be requested in it.

To summarize, the SNMP Manager determines the status of Agents by using the GET request to poll the Agents on a regular basis. However, if an Agent discovers an exceptional event, it will alert the Manager immediately, without waiting for a GET request, by sending a trap. The trap does not necessarily describe the problem in any detail; it simply informs the Manager that there is a change of state within the resources managed by the Agent, and the Manager can poll the Agent via an SNMP GET or GET NEXT request for more detail about the problem.


[ Top of Page | Previous Page | Next Page | Table of Contents | Index ]