Previous Table of Contents Next


Chapter 80
Using Intelligent Agents to Manage Multivendor Environments

Nathan J. Muller

The challenge of managing multivendor systems is increasingly being met through the use of intelligent agents, special software installed on managed stations on the local or wide area network that collects performance information in a standard format, implements preemptive actions based on predefined policies, and automates routine, repetitive tasks. These capabilities help minimize traffic on the network, increase systems availability and reliability, and reduce the overall cost of systems management.

INTRODUCTION

Today’s distributed computing environment usually includes a mix of UNIX, MVS, VMS, MS-DOS, NetWare, Windows, and other operating systems from multiple vendors. Managing a multivendor environment is difficult, especially because standards for systems management and underlying technologies are lacking. The OSF’s DME effort has failed largely because it attempted to address everything from applications to infrastructure all at once instead of setting priorities and proceeding incrementally.

Open systems management is compounded by the fact that every vendor defines the term differently and therefore takes a different approach to systems management. Some vendors offer integrated suites of applications; others merely provide platforms or frameworks and rely on third parties for specific applications that plug into them. Because vendors offer overlapping but not identical sets of products, users find product comparisons and selection difficult. In many cases, this difficulty has delayed full-scale implementation of systems management, especially for client/server environments.

The most promising approach to the challenge of managing multivendor environments is embodied in the concept of intelligent agents. Intelligent agents are special software installed on managed stations on the LAN/WAN that collects performance information in a standard format and implements preemptive actions based on predefined policies. Because filters and thresholds are used to send the management console specific items of information only, the software provides a more efficient process than continuous polling of every station to notify the central management station of all events. Using intelligent agents thus minimizes traffic on the network, which, in turn, consumes less bandwidth and contains costs. It also has the following business benefits:

  By detecting exceptional events and triggering automated actions to correct problems before they affect systems services, intelligent agents increase systems availability and reliability.
  By automating routine, repetitive tasks, intelligent agents relieve the systems administrator of a heavy work burden and reduce the overall cost of systems management. They speed up problem resolution while permitting businesses to get the most out of their technology investments.

Organizations that use intelligent agents to continuously monitor systems and applications for problems and carry out corrective actions on an automated basis thus achieve a competitive edge over companies that react to problems with manual processes.

KEY CONCEPTS IN SYSTEMS ADMINISTRATION

Manager and Agent

The key concepts in systems (and network) administration are the manager and agents. The manager is the workstation that is set up to view information collected by the agents. The agents are special programs that are designed to retrieve specific information from systems on the network. An application agent, for example, works on each workstation to log application usage. Workstation users are not aware of the agent and it has no effect on the performance of the workstation or the applications running on it. The collected information is retrieved from one or more databases for viewing on the systems administrator’s console.

The use of agents is not new, of course. The manager-agent relationship is intrinsic to most standard network management protocols, including SNMP. SNMP agents are now used routinely in all kinds of internetworking devices, including bridges, routers, hubs, multiplexers, and switches. Exhibit 1 depicts the interactions between an SNMP manager and agent.


Exhibit 1.  Interactions Between an SNMP Manager and Agent.

What is relatively new is the use of intelligent agents to manage systems-level activities. The difference between network-level (i.e., SNMP) agents and systems-level agents lies in their degree of intelligence.

Agent IQ

In the SNMP world, agents respond to polls from an SNMP management station that requests information on the operational status of network devices. Based on that information, agents are then directed by the management station to get more data, set management variables, or generate traps when significant events occur. By necessity, SNMP agents are dumb — that is, they act merely as passive data collectors that provide information only when they receive a request from the central management station. That such agents cannot perform filtering or processing on their own is acceptable, though, because bridges, routers, and other internetworking devices have limited memory and processing resources.

The idea of using SNMP to manage UNIX and other systems started to catch on a few years ago, but it quickly became apparent that dumb agents were not very efficient for this purpose. Dumb agents must be polled continuously by central management software, a process that increases network traffic. Today, many desktop systems reside at remote sites, and polling over wide-area links quickly drives up operating costs. Even on higher-bandwidth LANs, which are being increasingly burdened with multimedia and other delay-sensitive applications, traffic from continuous polling and the resultant data transfers can easily affect performance thresholds. Intelligent agents address these problems.

What makes intelligent agents so smart is the addition of programming code that tells the agent exactly what to do, how to do it, and when to do it. Because UNIX and other systems are equipped to run complex computations anyway, adding intelligent agent software does not significantly tax a machine’s resources.

CAPABILITIES OF INTELLIGENT AGENTS

The information collected by intelligent agents from multiple sets of data can be displayed in several ways (e.g., in cells, charts, and text), analyzed for such purposes as license management or inventory management, and printed as a detail or summary report. Agents are becoming an increasingly popular method of carrying out repetitive, programmable tasks. In essence, they play the dual role of manager and agent. Intelligent agents that reside on UNIX systems are capable of tracking hundreds of parameters. They are used mostly for scheduling, controlling, and monitoring the execution of processes across the network and ensuring their successful completion.

A comprehensive agent-based management application allows the systems administrator to perform the following main functions:

  View and manipulate network data.
  Automate file distribution.
  Maintain hardware inventory.
  Manage installed software, including application usage.
  Receive notification of network events.
  Establish and manage network printer support.
  Automate network processes, such as backup and virus detection.
  Monitor disk and file usage.
  Create task lists.
  Manage storage.

All of the agents that collect information in support of these functions are configured at the management console using commands selected from the menu bar. Once configured, each type of agent can be assigned an icon that launches its associated viewer for displaying collected information. The many types of agents available include those for monitoring disk, application, and file usage; distributing files; tracking network activity; and process scheduling. The following sections explain how intelligent agents work in the distributed environment and aid in multivendor systems management.


Previous Table of Contents Next

Copyright © CRC Press LLC