Version 6 (modified by Beth Plale, 14 years ago) (diff)


NetKarma Quarterly Report Q2 of 2010: 04/01-06/30/2010
PI: Beth Plale, School of Informatics and Computing, Indiana University Bloomington Co-PI: Chris Small, Global NOC, Indiana University

Quarter Accomplishments

The tangible accomplishment from Q2 is the GEC-8 NetKarma demo session poster that can be found at the URL below. Other accomplishments are of an ongoing nature and outlined in the next section.

Description of Work Performed During Quarter

After having successfully demoed collection and visualization of provenance from the Gush tool at GEC-7, in this quarter we pushed on several fronts:

  • Generalize the provenance collection mechanism – the adapter mechanism parses a log looking for provenance specific information. We are working on generalizing this.
  • Began the process of identifying provenance in RAVEN – we’ve looked over the available materials for RAVEN and have talked with the team, including a F2F meeting at GEC-8. This work is ongoing.
  • Refactor the provenance database schema to store provenance more efficiently.
  • Established involvement in the Instrumentation and Measurement Working Group. Attended the meeting June 8-9 in Chicago, IL. There are a couple areas where we have plugged in: in defining the archive repository, and in the measurement schema.

Project Participants Key participants in the NetKarma project for Q2 include: Beth Plale, PI Chris Small, Co-PI Girish Subramanian, PhD student – away at internship Yiming Sun, PhD student Devarshi Goshal, PhD student Sharanya Chinnusamy, MS student – graduated, at Microsoft David Ripley, technical staff Robert Ping, Project and Information Management

Publications & Documents

GEC-8 demo slides

PlanetLab Control Framework Cluster B meeting update slides


We continue to have discussions with the Raven project and expect to obtain a copy of the code in August 2010.

We met with the GMOC team and are studying the extent to which operational data collected by the GMOC, especially control plane related data, could be collected from the GMOC repository instead of the original source. The GMOC repository would provide a unified, normalized schema to query against compared to widely differing data and formats used by control plane and other data sources. Collecting data already obtained from various sources and stored in the GMOC database schema may speed and simplify the work need for data collection.

We had a preliminary discussion of how OpenFlow state such as flow tables and controller policy rules can be used. We focused on collection from the Indiana University OpenFlow campus trial infrastructure as an initial test but techniques should be applicable to all OpenFlow deployments.

Planned Activities for Q3

We talked with Vic Thomas, our GENI program manager, at GEC8 about collection at the Control Plane. It was thought that there were likely only a couple data items of interest at the control plane, and we should focus on working to collect those items more so than establishing deep connections to the tools, then quickly move to the instrumentation level where a provenance viewpoint and tooling can assist in making connections between an experiment and the measurement logs that are gathered. We have engaged in the I&M working group to explore the latter.

We will continue to explore provenance collection within Raven. We are developing a full distributed experiment, likely based on Bit Torrent, so as to have a richer testbed on which to work. The next step will be to collect from slices owned by others.