We present Flashback, a ready-to-use system for scalably handling large unexpected traffic spikes on web-sites. Unlike previous systems, our approach does not rely on any intermediate nodes to cache content. Instead, the clients (browsers) create a dynamic, self-scaling Peer-to-Peer (P2P) web-server that grows and shrinks according to the load. This approach translates into a challenging problem – a P2P data exchange protocol that can operate in churn rates where more than 90 % of peers can leave the overlay in under 10 seconds. This is atleast an order of magnitude higher churn rate than previously addressed research. Additionally, our system operates under two strict constraints – users are assured that they upload only as much as they download and second, end-user browsing experience is preserved, i.e., low latency downloads and zero configuration or download of any software. We believe these are very important for wide acceptance of the system. Various innovations were required to meet these challenges. Key among them are (a) A TCP-friendly, UDP protocol (Roulette) for Tit-For-Tat data exchange under extreme churn, (b) A novel data structure (NOIS) for partial-data management, (c) A distributed hole-punching protocol for automatic NAT traversal and (d) Automatic rendering of webpages using a technique we call the transported frame hack. Experimental results show the effectiveness and near optimal scaling of Flashback. For a webserver (and clients) running on a DSL-like connection, end-user latency increases only one second for every doubling in webserver load.

We present Flashback 1, a ready-to-use system for scalably handling large unexpected traffic spikes on web-sites. Unlike previous systems, our approach does not rely on any intermediate nodes to cache content. Instead, the clients (browsers) create a dynamic, self-scaling Peer-to-Peer (P2P) web-server that grows and shrinks according to the load. This approach translates into a challenging problem – a P2P data exchange protocol that can operate in churn rates where more than 90 % of peers can leave the overlay in under 10 seconds. This is at least an order of magnitude higher churn rate than previously addressed research. Additionally, our system operates under two strict constraints – users are assured that they upload only as much as they download and second, end-user browsing experience is preserved, i.e., low latency downloads and zero configuration or download of any software. Various innovations were required to meet these challenges. Key among them are (a) A TCP-friendly, UDP protocol (Roulette) for Tit-For-Tat data exchange under extreme churn, (b) A novel data structure (NOIS) for partial-data management and (c) A distributed hole-punching protocol for automatic NAT traversal. Experimental results show the effectiveness and near optimal scaling of Flashback. For a web-server (and clients) running on a DSL-like connection, end-user latency increases only one second for every doubling in web-server load. 1

In unstructured P2P systems, peers organize themselves into a random overlay. A challenging problem in these systems is to efficiently locate appropriate peers to answer a specific query. This paper proposes a semantic method in, which a query can be routed for appropriate peers instead of broadcasting or using random selection. This semantic is generally built from the contents of the peers, but can also bring in the implicit behavior of the users. The main objective of our method is to achieve better results in non-supervised tasks through the incorporation of usage data obtained from past search queries. This type of method allows us to discover the motivations of users when visiting a certain documents and peers. The terms used in past queries can provide a better choice of features queries. Hence, for each peer, our method learns from past queries to represent correlation between sent queries terms and related peers. We implemented the proposed method, and compared its routing effectiveness in terms of both recall and messages traffic with a broadcasting scheme (without learning). Experimental results show that our method is efficient and performs better than other non-semantic query routing methods with respect to accuracy. In addition, our approach improves the recall rate nearly 90 % while reducing message traffic dramatically compared with Gnutella protocol.

I hereby declare that all information in this document has been ob-tained and presented in accordance with academic rules and ethical conduct. I also declare that, as required by these rules and conduct, I have fully cited and referenced all material and results that are not original to this work. Name, Lastname: Signature: iii