Version 1 (modified by Aaron Falk, 13 years ago) (diff)


ProtoGENI Project Quarterly Report for 4Q 2008

I. Major accomplishments

A. Milestones achieved

Year 1(g): Gave a demo of progress early, at GEC #3

B. Deliverables made

Documentation of design decisions and plans up at

Release of Emulab software, including early versions of ProtoGENI extensions, made under the AGPLv3 in July (before contract awarded)

Early RSpec prototype (still in progress) released at:

Early credential and ticket formats (still in progress) released at:

II. Description of work performed during last quarter

A. Activities and findings

In our role as a control framework, most of the activity billable this quarter to this contract has been integration, collaboration (see below), and running of a clearinghouse and aggregate component managers. We have brought up a clearinghouse and two aggregate component managers at Utah. These are by no means complete, but represent a strong start. We have lined up four external sites to host ProtoGENI edge nodes. Two of these sites have been running for over a moth, and the other two will come up after the holidays.

Under separate NSF funding, we have done extensive development on ProtoGENI, including:

Work on the aggregate component manager frontend to the Emulab management software

Building software to run a clearinghouse

Building software to create and validate user identities, credentials, and tickets

Enabling federation between sites running Emulab software, using the ProtoGENI APIs

The last is particularly important, as it will be a primary method for integration with other cluster members who run Emulab instances (which is nearly all of them.) An early version of this federation (of aggregates) has been tested within Utah, using separate Emulab instances, but remains to be deployed at other sites. Our strategy is push for early adoption of the federation software, before it is fully complete, so that its development can be guided by lessons learned during deployment.

We participated in GEC3, giving a talk and a demo, and had discussions with several other projects while there. We also attended the GENI-FIRE meeting in Madrid, giving a talk and having discussions with possible international collaborators.

B. Project participants

University of Utah

Subcontracts for HP and Internet2 are pending, as Utah's contract with BBN was executed only in December.

C. Publications (individual and organizational)

We presented a poster at OSDI 2008, accompanied by a demo (the same one given at GEC #3). Poster and demo were well-received.

Title: ProtoGENI: A Network for Next-Generation Internet Research

Authors: Robert Ricci, Jay Lepreau, Leigh Stoller, Mike Hibler, and David Johnson

D. Outreach activities

Our poster and demo at OSDI gave us an opportunity to discuss GENI with a number of people involved in the distributed systems community, many of which weren't yet familiar with GENI, or who had questions regarding it. We also participated in the GPO's outreach activity to the European community in Madrid.

E. Collaborations

Beginning in early October, we have organized bi-weekly conference calls with the members of our cluster, in order to foster collaboration within it. Participants have included all groups initially assigned to our cluster, our subcontractors, the Programmable Edge Node project, and the Security Architecture project.

We have interacted with a number of other GPO-funded projects within and outside of our cluster about collaborations and integrations, including:

BGP Mux: Held a discussion with this group and others about how the virtualization work they are doing can be used by our project and connected to the Programmable Edge Node project. We are in the process of looking at the implementation they have thus far. Georgia Tech also now hosts two ProtoGENI edge nodes.

CMU Testbeds: Providing code/software support to this project as they come up to speed on the Emulab and ProtoGENI software for deployment of HomeNet.

Instrumentation Tools: They have given a demo of their existing work to one of our cluster conference calls, and we have begun talks about the specific tasks needed to integrate their work into ProtoGENI. The University of Kentucky will soon host two ProtoGENI edge nodes.

Measurement System: Have had discussions about the specifics of how integration with this project will work, and running the necessary Emulab code to join our clearinghouse. Also discussed possible low-layer connectivity between our projects. The University of Wisconsin-Madison now hosts two ProtoGENI edge nodes.

Enterprise GENI: Exploring the idea of providing OpenFlow capabilities on the ProtoGENI backbone

SPP Overlay Nodes: Considering sharing the bandwidth provided by Internet2 with this project, using the switches to be purchased under our contract. We have also had discussions with the GPO and Internet2 regarding this arrangement.

Million-Node GENI: Working on making this project available in a controlled environment, and running it on ProtoGENI nodes.

Programmable Edge Node: This project has joined our cluster, and we have had talks about how integrate this project with the work that we and others in our cluster are doing.

GMOC: Initial talks about providing emergency shutdown and operational information

Security Architecture: This project has been a regular participant in our cluster conference calls.

DETER: Initial contact regarding the Deter Federation Daemon, and possible low-layer interconnect between the clusters

PlanetLab: We have been in occasional contact with this project, with the goal of reducing the differences between the APIs provided by these two control frameworks

We have also been in close contact with the GPO, including a demo of Emulab and ProtoGENI to the GPO technical staff. We responded to the Substrate Working Group's call for information regarding hardware to be deployed, and also posted this information on . We have been working with Internet2 and our campus OIT to plan the backbone to de deployed under this subcontract. We have been in close contact with our other subcontractor, HP, regarding their role of providing switches and hosting an edge cluster, and they will soon host two ProtoGENI edge nodes. Finally, we have held preliminary discussions with international Emulab sites about possible international federation efforts.

F. Other Contributions