wiki:InfoSubstrate-3Q10-status

Version 1 (modified by sollins@csail.mit.edu, 9 years ago) (diff)

--

InfoSubstrate Project Status Report

Period: June 30, 2010 through November 5, 2010

I. Major accomplishments

In this period, we accomplished completion of the first of the two papers under this contract. The first paper addresses the architectural challenges of providing an information substrate for federated network management in GENI. It is available on the wiki page of this project.

A. Milestones achieved

In this period we completed both S2a and S2b on time.

B. Deliverables made

The deliverables for this period were the first and final drafts of the paper on architectural challenges or design goals for an information substrate for federated network management. Both versions of the paper were delivered, by being posted on the project wiki.

II. Description of work performed during last quarter

During this period, more background reading was performed as well as further interviews. The PI wrote both the preliminary and final versions of the first paper of this contract. In addition, the PI has attended GEC9 on a GENI travel grant. This provided a valuable opportunity to interact across a number of the working groups and clusters. Of particular interest are COMIS, Instruementation and Measurement, and the Control Framework. The Instrumentation and Measurement WG appear to be very focussed on their narrow problem of providing a capability for experiments to be instrumented and measured, but less so with respect to the broader set of resources. There were two ways in which the interaction with them was disappinting. First, they did not want to think about the fact that if something is not working correctly in an experiment, that diagnosing that problem will need to span the spectrum from the experiment itself to the various components of the GENI infrastructure. Such problems can easily be more complex than can be diagnosed only from the "experiment" perspective. In addition, they did not want to think about the fact that managing a GENI aggregate may need many of the same facilities as those they are creating for the experiment. The COMIS and Control Framework Working Groups are more interested in the kinds of challenges we are identifying in our work in this project.

A. Activities and findings

The key, perhaps novel, insight identified by the paper is that there are two sets of orthogonal forces that must be considered in network management. The first is that management issues may be span multiple network architecture layers, so there is a vertical aspect to integrated network management. The second is that federation provides a horizontal architectural challenge. A problem or issue requiring management may on derive from the fact that peers are interacting. These simultaneous challenges are what make network management for GENI require a more complex management architecture than exists traditionally.

B. Project participants

Karen Sollins

C. Publications (individual and organizational)

None

D. Outreach activities

Several of my students are working on a multi-domain network management framework more generally. GENI is providing a concrete example in their work, which otherwise will be more general and more far-reaching.

E. Collaborations

None

F. Other Contributions

None