Netflow meeting 20060112

JRA1 D3.4 Flow Monitoring - minutes

Date/time: 12th January 2006.

Ad-hoc meeting during GN2 Technical Workshop in Cambridge

Attendees: Hans Trompert (HT), Jon Kare Hellen (JH), Olav Kvittem (OK), Igor Velimirovic (IV)

There is a task of integrating netflow measurement infrastructure (netflow MPs, netflow server) into overall PerfSonar architecture. At this point it is not clear whether we should implement netflow as 'standard' MA, or it should act as a MP (measurement point), or maybe act as an independent module within framework. It is obvious that due to the size of netflow data, implementation should consider netflow specific issues. HT informed that there is a task of implementing 'stream data export' within perfSonar, he will check the details.
It is obvious that we should consider aggregation and anomysation issues in order to:
- provede aggregated data (due to size of netflow data)
- provide some 'anonymized' data to the users, for example with hidden IP addresses
Group strongly feels that netflow version 9 or/and IPFIX data format should be considered in this research activity as a priority. Therefore tools like flow-tools, for which it is almost certain that will not support those flow formats, could not be selected solution.
Since some implementation of netflow collector tool provide data in rrd format, it is possible to create perfSonar MA from that. OK also reported that STAGER team is going to prepare a MA based on STAGER data.

Actions

IV to contact GARR, and check whether we should continue with flow-tools.
IV to check with Internet 2 (Abilane) on how they perform the netflow data anomysation.
OK to check with Stager developers is there a time plan for implementing MA using Stager.
All: in all contacts with equipment and software manufactures check the status of the IPFIX flow format implementation.