Summary

This chapter introduced the idea of policy-based management as a means of dealing with demands for service management in a complex environment with tight time constraints. Automating many of the responses and procedures minimizes staff labor, reduces staff mistakes, and provides the speed needed to meet stringent SLA compliance criteria.

Policies serve two main purposes: they define what actions the management system takes in certain situations, and they prohibit other management activities that are irrelevant to a specific problem.

Policy systems have enforcers to determine the appropriate actions to take on service flows. Policies are distributed using push, pull, or hybrid approaches. The push model is very effective for abruptly changing the policy system behavior. The pull model enables each component to ask for information as needed.

Policies for services management evolve by automatically integrating more functionality. Consider the policies that could be activated when a desktop initiates a streaming connection. Collectors inside the desktop are activated to measure the latency and packet loss on the connection. Alerts are forwarded if the measurements indicate an actual or potential service disruption. Monitoring is discontinued when the connection is terminated.

A security breach might activate a set of policies that adjust firewalls, isolate key resources, inform corporate management, and track the intruder while alerting the management team.

Policy-based products for service management are still maturing and administrators need to assess their actual capabilities carefully.

For any policy, the output of a decision is only as good as the quality of the input. In selecting where to apply policy-based management, as much consideration must be given to the information used to make the decision as to the automation of possible outcomes. This reinforces the importance of good instrumentation and event management for good policy-based management.