IRTF P2PRG Content, Resource, Service Discovery Subgroup

IRTF P2PRG Content, Resource and Service Discovery (CORE) Subgroup

Chair(s):

P2P Research Chair(s):

Mailing Lists

The email list is p2prg@irtf.org. You need not be a list member to send mail to the list. To subscribe, visit the P2PRG mail page or send an email to p2prg-request@irtf.org.

An archive of the email list is available at P2PRG mail archive.

Description

Discovery of content, resources and services is a fundamental operation for peers in a peer-to-peer network. Many discovery mechanisms have been proposed. However existing designs have limitations in one or more areas of scalability, security, interoperability, or generality. The purpose of this subgroup activity is to define a research agenda within the P2PRG community to evaluate existing research, identify requirements, and develop solutions for wide-area peer-to-peer content, resource, and service discovery.

We expect to address this goal in two dimensions:

An evolution of existing Internet server-based infrastructure such as DNS and SLP. Examples of recent research results that are starting points include DDNS and OSDA.

A general-purpose Internet mechanism on which various wide-area P2P applications can be constructed. Examples of research systems include INS/Twine and OpenDHT.

We expect scalability, security, trust, interoperability, performance, and generality to be important requirements in the overall architecture.

Existing methods include client-server mechansisms (such as SLP), mechanisms integrated into device network adapters (such as Bluetooth SDP), mechanisms integrated in middleware platforms (such as Web services, Jini, and JXTA), and home network oriented solutions (UPnP SSDP).

In addition there are experimental systems for service discovery such as SSDS and these extensions to SLP: WASRV, wide-area SLP, and mesh-enhanced SLP; and the use of wide-area multicast to advertise and discover services.

Further there are various peer-to-peer systems and proposed designs, both unstructured and structured by which large numbers of peers share content and services. Typically large-scale unstructured peer-to-peer systems use limited scope flooding to reach other peers with requests, whereas structured peer-to-peer systems use a distributed hash table (DHT) to store key-value peers.

Associated with such discovery mechanisms are related mechanisms for service/resource description, advertisement, notification, and service invocation.

Goals and Milestones:

Problem statement. There may be separate problem statements for the Internet infrastructure case vs the application infrastructure

Survey of related work. Several surveys exist, but do not specifically address the two goals above.

Experimental plan. We envision development of a large scale testbed for experimenting with wide-area service discovery. We believe a collaborative testbed is a unique benefit of working in the IRTF versus other research contexts. The experimental plan will address the resources, requirements, and participants needed to launch this.

Experimental results. Dependent on level of participation of the P2PRG community, summary of experiments related to comparing designs and evaulating them with respect to functionality, scalability, performance, and scalability.

Internet-Drafts:

P2PRG CORE Subgroup Problem Statement problem statement id Jan 2006.
Tools for Peer-to-Peer Network Simulation. draft Jan 2006.

Requests for Comments:

Previous Meeting Material:

Presentations at IETF 64 (Nov 05, Vancouver)

Overview of P2P Overlay Design (presented at P2P-SIP Ad Hoc at IETF 64). slides Nov 2005.
Overview of P2P Overlay Design (updated). slides Dec 2005.

Presentations at IETF 65 (March 06, Dallas)

CORE subgroup problem statement John Buford, Panasonic Princeton Lab
Survey of P2P Simulators Mario Kolberg, Stirling U.
Simulating P2P middlware with AgentJ Ian Downard, NRL
TCP Relay Selection Keith Ross, Polytechnic
DDOS Attack Keith Ross, Polytechnic