Search on the Cloud File System

Rodrigo Savage, Dulce Tania Nava, Norma Elva Chávez, and Norma Saiph Savage

Keywords

Distributed Computing, PeerToPeer Computing, Heterogeneous Computing, Algorithms For Heterogeneous Systems, Distributed Shared Memory

Abstract

Research in peer-to-peer file sharing systems has focused on tackling the design constraints encountered in distributed systems, while little attention has been devoted to the user experience: these systems always assume the user knows the public key of the file they are searching. Yet average users rarely even apprehend that file public keys exist. File sharing systems which do consider the user experience and allow users to search for files by their name, generally present centralized control and they show several severe vulnerabilities, that make the system unreliable and insecure. The purpose of this investigation is to design a more complete distributed file sharing system that is not only trustable, scalable and secure, but also leverages the user's cognitive workload. We present a novel algorithm that by mining a file's information designates relevant keywords for the file automatically. These keywords are later utilized for the file search and retrieval. We also designed a metric for assigning relevancy to the files retrieved in a search, bettering the search results. We also create a modern mechanism for enabling file searches based on categories. Search on the Cloud is built on Pastry. Our system integrates these components, as well as good design principals from previous distributed file sharing systems to offer a trustable, scalable, secure and novel distributed file sharing system that an average user could utilize for file search. Our system is named “Search on the Cloud”. The novelty of our approach is that our system provides an intuitive search modality, while still preserving an entirely distributed approach.

Important Links:



Go Back