How does PfR Manager work?
PfR Manager enables users to easily define, implement and monitor traffic classes and associated policies to dramatically improve network performance and operating cost.
Traffic Classes:
The first step when implementing PfR and PfR Manager is to define traffic classes. Traffic classes can be operator-defined as a combination of IP addresses, port numbers, protocols, and DSCP values. Examples of traffic classes include VoIP, web traffic, email and key business applications. With PfR Manager, the use of drop down menus, text boxes and mouse clicks makes this step easy.
Monitoring Type:
Next, users may choose their monitoring types. Operators can select between active monitoring, passive monitoring or both, which is the default setting. Active monitoring occurs via IPSLA synthetic traffic generated by the border routers. Active monitoring provides data on delay, reachability, jitter and MOS. Passive monitoring occurs at the router as well via NetFlow data collection. It provides network measurements of delay, packet loss, reachability and throughput. In addition, PfR Manager can be configured to monitor link usage and load distribution.
Policies:
Policy configuration is the final step. Traffic class policies will be based on the metrics that are enabled through passive and/or active monitoring. There are two categories of traffic class policies: threshold and security.
Threshold policies will enable dynamic responses to network conditions that vary from policy settings. “Out of Policy” conditions result from a deviation from recent values or absolute values, dictated by relative or absolute policies. A relative policy requires that a traffic class metric not vary by a given percentage. For example, the delay for a retail store location should not increase more than 40% to remain “In Policy”. An absolute policy requires that a traffic class metric remain within an absolute value. For example, VoIP traffic MOS scores cannot fall below 3.7. A final element to threshold policies is a choice between “Better” and “Best”. Selecting “Better” performance for a policy means that PfR will find alternatives to keep the traffic class “In Policy”. Choosing “Best” mandates that PfR continually shift a given traffic class to the link that optimizes the selected policy metrics.
Security policies enable effective handling of suspicious or malicious traffic. Suspicious traffic classes can be defined and policies established that capture such traffic in a “sinkhole” for further evaluation. Known malicious traffic classes can be defined and then discarded immediately into a “black-hole”.
 Traffic classes and associated policies are easily created and administered through the graphical user interface of PfR Manager. Observe and Control:
If PfR Manager is run in “Observe” mode, then no network routing changes will be made, yet network performance can be viewed and recorded. Baseline reports also will be generated. When PfR Manager is activated in “Control” mode, it then both observes and takes action as required to bring traffic classes back to “In Policy” status.
Reporting:
PfR Manager provides graphical and tabular reports detailing the behavior of the network as well as the changes taken by PfR itself. Graphical reports represent current performance against established policies. PfR Manager will display which traffic classes are in and out of policy in real time as well as in historical reports of performance across all enabled metrics. Any policy failures and resulting routing changes will be recorded and displayed in the Event History. Exit links are monitored and their performance recorded for future display and reporting. Extended reporting capabilities are also built into PfR Manager, which allow for filtering and customization for specific audiences.
 Graphical report showing historical traffic class performance and associated route changes. |
