wiki:sppReport-3Q09

Version 1 (modified by Christopher Small, 14 years ago) (diff)

--

Internet Scale Overlay Hosting Progress Report 6/1/2009-9/30/2009)

1. Project Activities this Quarter

Installation and Checkout of First Three SPP Nodes

We completed preparations for the deployment of our first three Superchaged Planetlab Platforms, which were installed this quarter in Washington DC, Kansas City and Salt Lake City. There were several minor installation issues that were easily resolved. There was also one larger issue that required the replacement of one of the Radisys 7010 Network Processor blades in the Salt Lake City node, with a spare. It appears that the power supply module on the board has failed. We plan to return it to Radisys for repair.

All three nodes have been checked out and are now operating correctly. There are some minor issues that should be addressed when there are personnel visiting the sites for other reasons, but none that require a separate visit.

Initial Planning for Demo at GEC 6

We are planning to do a demo using the deployed SPP nodes at GEC6. We will use standard PlanetLab nodes to generate traffic carried by the experimental networks hosted by the SPPs. We plan two demonstrations. The first will involve IPv4 routers with a fastpath that is mapped onto the NPEs. The second will involve an early prototype of an overlay network architecture called Forest, which is designed to support a class of real-time distributed applications (more details below). The preparation for the demo is now underway, and will consume a a major portion of our effort in the fourth quarter.

System Architecture Document

This document provides a detailed description of the hardware and software components used to implement the SPP. It is now available on the project wiki page.

User Documentation

Work continues on a wiki that will provide complete user documentation on how to use SPP nodes to setup and operate experimental networks. We plan to have a preliminary version of the wiki available by GEC 6. This will include a detailed description of the demo planned for GEC 6.

Flow Monitoring

The node installation work has dominated our activities this quarter, so we have made little progress on the the flow monitoring subsystem needed to keep track of outgoing Internet traffic. This subsystem will allow us to identify and disable experiments running on the SPPs that send unwanted traffic that triggers complaints. This involves data plane software in the egress side of the Line Card, data collection software in the Line Card's xScale management processor, and data aggregation and archiving software running in the CP. Once this is complete, it will be possible to access the outgoing flow statistics through standard PlanetLab management mechanisms. This will need to be completed before we can open up the SPPs for general use.

Development of Second Version of Network Processor Datapath Software

Work is continuing on the development of a more complete version of the datapath software for the Network Processor Engine (NPE) used to implement slice fastpaths. This new version is needed to support full 10 Gb/s operation and multicast packet delivery. We have not been able to devote a lot of resources to this effort so far, but expect to give it more attention after the initial deployment is complete.

Preliminary Work New Fast Path Code Option

We have decided to take a different approach to the development of the next fastpath code option. We have designed a new network architecture called Forest, which is designed to effectively support a particular class of distributed applications that includes distributed interactive simulations, multiplayer games and virtual worlds. The common element of these applications is that they involve many distributed components that issue periodic reports that other components may subscribe to. Subscriptions are highly dynamic and reports must be delivered to their subscribers in a non-stop fashion, regardless of the specific traffic pattern. The central primitive in Forest is a tree-structured communications channel called a comtree. Each distributed application session is assigned its own comtree, which is routed and provisioned to meet the needs of that session. We will demonstrate an early version of Forest that runs on the SPPs' conventional server blades at GEC6. We plan to develop a fast path based on this implementation, that can be mapped to a new NPE code option.

Bug Fixes

Tracking down and fixing bugs is a continuing process in a project like this. There have been no major new issues this quarter, but this continues to occupy a signficant amount of staff time.

Transition of Staff Responsibilities

2. Milestones achieved

Installation of SPP nodes at Washington DC, Kansas City and Salt Lake City.

Completion of System Architecture Document

3. Deliverables made

none yet

4. Project participants

Jon Turner - PI
Patrick Crowley - PI
John Dehart - technical staff
Dave Zar - technical staff
Ken Wong - technical staff
Mike Wilson - graduate student
Mart Haitjema - graduate student
Ritun Patney - graduate student

5. Publications (individual and organizational)

None yet.

6. Outreach activities

None yet.

7. Collaborations

Have had some discussions with Larry Peterson of Princeton, on rspecs for the SPP.