Browse Prior Art Database

Catalog Caching to improve Search Efficiency in distributed Peer-to-Peer file sharing systems.

IP.com Disclosure Number: IPCOM000014991D
Original Publication Date: 2001-Sep-01
Included in the Prior Art Database: 2003-Jun-20
Document File: 2 page(s) / 43K

Publishing Venue

IBM

Abstract

This is a technique to improve the efficiency of searching for files in a decentralized file sharing system. There are two major methods of peer-to-peer file sharing. The first is via the Centralized Server method of P2P file sharing. In this method the Central Server will track the locations of any files kept by the clients in a centralized index that can be queried by all attached clients. Examples of this include Napster and Scour Exchange. The other method is the Decentralized Method of File Sharing. In the decentralized model the member is both a server and a client (Gnutella refers to each machine as a servant). Each servant is responsible for keeping track of its own files. It will query its "peer" for a file. Each peer will then ask its peers, propagating to each member of the system until the request's time to live requirement is satisfied. The problem with the Decentralized Method of File Sharing is slow results to seaches. Delays are created as queries are passed from servant to servant. My idea improves on the Gnutella Protocol to decrease search times by caching information locally. I accomplish this by creating a method to share catalogs between servants, and allowing the searching of peers to happen on the local machine, reducing the amount of delay in a search. Catalogs are the key to this scheme. A Catalog has two parts, a CatalogHead and a CatalogBody. They are composed of the following: CatalogHead: Catalog Owner

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 2

  Catalog Caching to improve Search Efficiency in distributed Peer-to-Peer file sharing systems.

    This is a technique to improve the efficiency of searching for files in a decentralized file sharing system. There are two major methods of peer-to-peer file sharing. The first is via the Centralized Server method of P2P file sharing. In this method the Central Server will track the locations of any files kept by the clients in a centralized index that can be queried by all attached clients. Examples of this include Napster and Scour Exchange. The other method is the Decentralized Method of File Sharing. In the decentralized model the member is both a server and a client (Gnutella refers to each machine as a servant). Each servant is responsible for keeping track of its own files. It will query its "peer" for a file. Each peer will then ask its peers, propagating to each member of the system until the request's time to live requirement is satisfied. The problem with the Decentralized Method of File Sharing is slow results to seaches. Delays are created as queries are passed from servant to servant. My idea improves on the Gnutella Protocol to decrease search times by caching information locally. I accomplish this by creating a method to share catalogs between servants, and allowing the searching of peers to happen on the local machine, reducing the amount of delay in a search.

Catalogs are the key to this scheme. A Catalog has two parts, a CatalogHead and a CatalogBody. They are composed of the following:

CatalogHead:

- Catalog Owner

- Live Date

- Connection Type

- Connection stability (Dynamic vs Static IP)

- PingPath :

- Size of catalog

CatalogBody:

- File names

- File ID's

- File Sizes

When sharing a catalog, the origin machine will send the Catalog Head first and wait for the requesting machine to ask for the second part. This allows the requesting machine to decide if it is interested in the new Catalog. Requesting machines should be capable of filtering which ca...