Performance Measurement and Monitoring

All research and education networks make use of a range of tools for monitoring the status of their network infrastructure, from simple data on the level of link usage to detailed protocol analysis. However, the methods used for collecting and analysing data vary widely between networks. The definition of a user is similarly varied, and different policies concerning access to the available data are in force in different countries.

Furthermore, in any single international path between two end points, at least five networks are involved:

Two university or campus networks
Two national backbones (the participating NRENs)
The European backbone (GÉANT or GÉANT2).

An incident that disrupts the data flow could occur anywhere in these five domains, and as a result, can currently be extremely difficult to identify and remedy. Increased research activity in this area is therefore essential, to support best practice in delivering end-to-end service of a high quality to end users of research and education networks in Europe.

In order to provide users with reliable and consistent performance monitoring data from multiple networks, the Performance Measurement and Monitoring research activity will develop a comprehensive network monitoring facility. This is a key aspect of the project. The facility will provide access to monitoring data for all those involved in the end-to-end management chain.

The Performance research activity has three main goals:

To provide users (of many types) with network management and performance information, through a specified interface common to all the networks and through visualisation interfaces. User communities may include:

Staff in network operations centres (NOCs) for campus, regional and national networks, and for GÉANT2
The Performance Enhancement and Response Team (PERT)
High-volume users
End users with an interest in the development and behaviour of R&E networks.

To widen the set of performance data that is collected. The basic dataset will continue to be widened over the lifetime of the project beyond the initial range of metrics identified. To monitor the metrics of interest, some existing tools will be modified to suit the needs of the activity. Some examples of initially-identified metrics include:

Link usage
One-way delay
Packet loss
Delay variation
Achieved data rates.

To design and implement tools for collecting and processing network measurement data belonging to multiple networks. This should be independent of the various measurement systems employed, and interoperable with all. It manages the presentation of data to the end users, tailored according to their profile.

The activity recognises the need to provide different granularities of data according to user type. This is vital for performance measurement, but especially for network management; precise and granular data is imperative for ensuring optimum application tuning, and successful and efficient troubleshooting.

Click here to view the perfSONAR MDM release 3.0 - Product Brief