A Parallel Download Protocol for Internet-based Distributed Systems

摘要

Parallel transfer of large amounts of data is attractive as (ⅰ) the data can be received in less time and (ⅱ) is more resilient to connection failures. This is often used in e-Science under the field of data sharing. Data from one research project should be easily assessable to others regardless of the size of the data thus parallelism is used to help share the data quickly. However, there are many tasks to be carried out to make it possible: replica discovery and load detection just to name two. A common problem with most solutions is they overlook tasks such as load detection. PHTTP is an innovative protocol which handles the tasks behind parallel transfer while presenting itself as a normal sequential transfer. The novelty of PHTTP is it combining both the distributed nature of BitTorrent and the commonly used HTTP protocol.