pWeb: A distributed web hosting infrastructure

集中式的Web托管对于今天的Web来说是不够的，因为Web被密集地用于托管大量的多媒体内容。集中式技术有几个固有的缺陷，包括：可伸缩性问题；单点故障；管理开销；以及托管费用。这些问题可以通过在具有网络和存储能力的消费设备的对等(P2P)网络上分发网站和大量网络内容来解决。由于同行将协作托管每个网站(或其中的一部分)，因此对网站的并发访问没有限制，没有单点故障，并减少了对昂贵的高端Web服务器的依赖。这还将减少服务器场的管理开销和电力消耗。P2P网络托管之所以具有挑战性，有两个原因：第一，与Web服务器相比，节点的正常运行时间要短得多；第二，与固定网站相比，共享内容在P2P网络中频繁更改位置。这些动态行为带来了以下研究挑战，我们的目标是解决：(A)命名：Web命名方案，统一资源定位符(URL)，不适合P2P环境，因为URL是位置相关的，并且URL分配需要中央命名机构；(B)索引和搜索：健壮和高效的索引对于跟踪不断变化的对等点和内容的位置是必不可少的；(C)可用性：为了确保P2P环境中网站的全天候可用性，我们提出了一种基于时间的复制策略，复制开销最小；(D)对等体系结构：我们提出了一个蓝图，为常规对等体系设计一个轻量级的网络托管体系结构，运行在管理要求最低的低端、支持互联网的家庭网络设备上。该方案提出的P2P虚拟主机框架新颖而全面。我们不仅期望为上述研究问题设计出创新的解决方案，而且相信这些发现将为学术界的继续研究开辟新的可能性。此外，在这个项目中产生的P2P客户端实现可以被加拿大的IT行业采用，作为实现P2P Web托管的基础。

Centralized web hosting is not sufficient for today's Web, which is intensively used for hosting voluminous multimedia content. Centralized technology has several inherent deficiencies, including: scalability issues; a single point of failure; administrative overhead; and hosting expenses. These problems can be addressed by distributing the websites and voluminous web contents over a peer-to-peer (P2P) network of consumer devices with network and storage capabilities. As peers will be hosting each website (or a part of it) collaboratively, there would be no limit on concurrent access to a website, no single point of failure, and a reduced dependency on expensive high-end web servers. This would also reduce administration overhead and power consumption at server farms. P2P Web hosting is challenging due to two reasons: firstly, uptime of a peer is much shorter compared to a web server, and secondly, a shared content frequently changes location in a P2P network, in contrast to stationary websites. These dynamic behaviors can raise the following research challenges, which we aim to address: (a) Naming: the Web naming scheme, Uniform Resource Locators (URL), is not suitable for P2P environment as URLs are location dependent and URL assignment require central naming authority; (b) Indexing and searching: robust and efficient indexing is essential for tracking continuously changing peers' and contents' locations; (c) Availability: to ensure 24/7 availability of websites in P2P environment we propose a time-based replication strategy with minimal replication overhead; (d) Peer architecture: we propose a blueprint for devising a lightweight web hosting architecture for regular peers, running on low-end, Internet-ready home-networking devices with minimal administration requirements. The P2P Web hosting framework presented in this proposal is novel and comprehensive. We not only expect to devise innovative solutions for the above research issues, but believe that these findings will open new possibilities for continuing research within academia. Moreover, the P2P client implementation to be produced within this project can be adopted by the Canadian IT industry as a basis for bringing P2P Web hosting into reality.