February 14, 2005

SPEAKERS
Eric Boyd, Internet2 [PPT]


Eric provided an update on the piPEs project. He gave a brief overview of the effort for folks who haven’t heard of the project at previous meetings. He noted that the project approach is akin to Metcalf’s Law in which the value of a single measurement node is increased, proportional to the number of measurement nodes within a network.

Goal 1 is development of a federation of measurement frameworks – working with the Global Grid Forum’s Network Measurement Working Group (GGF NMWG) to develop a common language for measurement frameworks (standards) so that each framework is interoperable with other existing frameworks and, as new frameworks (and tools and applications) are developed, the value of being interoperable makes using standards more useful. Developing with GGF NMWG a schema for test requests and responses that can be used in tool and framework development – Version 1.0 is currently out; Version 2.0 is due out soon. (Note: this WG is meeting on February 16, 2005 from 2-6 p.m. at the hotel.)

Goal 2 is to actually develop an interoperable measurement framework prototype – working with the developers of GÉANT2 measurement infrastructure group (JRA1) to create a framework that can be adapted by both piPEs and JRA1. A whitepaper on the architecture is available (from the piPEs web page) and the group has begun to work on the draft prototype architecture that will be completed by June.

Federation tools currently under development include three open source tools that are being fairly widely adopted at this point (though not all by the same location) – BWCTL, NDT, OWAMP. The difficult part, at this point, is the development of a lookup service; we are currently hand-maintaining a directory of known and shared measurement points.

The primary goals for the next year is to expand this list, complete the prototype that will be the next generation of piPEs framework as well as the GÉANT 2 framework, and to begin holding deployment workshops at gigaPoPs to get NOCs and any of their connectors who are interested, up and running (hands on – participants leave with the framework and tools installed on at least one location on each “campus”, most likely the edge router) and an understanding of how to employ the framework.

The biggest barrier to development: lack of resources! Internet2 has invested some resources into the project and so has the European Commission. We have been working with GÉANT 2 JRA1 team because we have common goals and they have funding to deploy a measurement framework. Both groups are actively involve din the GGF NMWG, we are evaluating each other’s tools, and we’ve held several workshops and demos of interoperability of existing tools/frameworks (Caltech to CERN, for example).

The best way to achieve our goals was to develop a single framework (instead of two different interoperable ones) – we now have a set of goals, a whitepaper outlining the architecture, and have begun work on the prototype. Missing pieces at this point – focus of work – is on lookup/discovery of nodes and incorporation of AA (security) components.

The current design emphasizes local control, services-orient measurement framework, which is dynamic (allowing nodes to come up and leave without affecting the value of the system unduly). Services: lookup/discovery, authentication, MP, measurement archive, resource protector, transformation (topology, etc.). He indicated that the piPEfitters BoF (Tuesday, February 15, 2005) would cover more details of the next generation of the piPEs software.

This is an open source, shared development, common service-based architecture – we are looking for more deployment and need input on what end users/NOCs/network administrators are looking for so that the development can take this into account.

For more information on piPEs, see http://e2epi.internet2.edu/e2epipes/.