Professional Joomla Templates Free

Efficient BitTorrent-like Content Distribution for Cloud Storage Services

Research Area: Peer-to-peer computing
Status: Finished Degree: Phd
Directors: Students:
Proposed start date: 2013-03-01 Proposed end date: 2017-01-17
Digital version

With the ever increasing Internet traffic, peer-to-peer (P2P) content distribution has emerged as an alternative to the traditional client-server model, especially with the recent bandwidth soar on the edges of the Internet. Data centers with limited bandwidth budget can benefit from the upload speed of the clients interested in the same content to improve the overall Quality of Service (QoS). This can be done by introducing a P2P protocol, BitTorrent for instance, when the load on a certain content becomes high.

The main challenge is to decide when is the best time to switch from the classic distribution protocol (HTTP) to BitTorrent, for each requested file. In fact, it is commonly assumed that BitTorrent is only efficient with big files and large sets of users, and that client-server protocols (including HTTP) perform better in the distribution of small files. However, to the best of our knowledge, there is no concrete analytic analysis of the transition point in efficiency between BitTorrent and HTTP. Our first contribution consists in the investigation of this transition point and the proposal of an algorithm that provides a simple switching strategy based on the QoS requirements of the system.

As a result from the introduction of BitTorrent, the data center will be faced with the challenge of managing its resources among clients that can be using different download protocols (HTTP and BitTorrent). To this extent, we calculate the amount of data center bandwidth needed to ensure a given ratio between the download times in HTTP and BitTorrent and propose a bandwidth allocation algorithm that decides the most suitable protocol for each case and provides the corresponding bandwidth allocations at the swarm level.

Nevertheless, the benefits that can be derived from the introduction of BitTorrent are tied with the number of simultaneous download requests of the same content. This number can be very limited in comparison with the number of separate download requests related to different files.To increase the contribution of the clients and benefit from the resources of peers involved in separate downloads, it is possible to make clients from different swarms cooperate with their spare bandwidth. We propose cross-swarm bundling to mitigate the problem of lack of simultaneous download requests. Cross-swarm bundling is the process of merging two swarms together into a single one. We prove that this technique can have a positive effect on the Quality of Experience (QoE), and present a methodology to implement bundling in data centers based on graph matching techniques.

All our proposals are evaluated using a trace of a Personal Cloud (Ubuntu One), and the results show that BitTorrent is can improve the performance of these systems when the cloud’s available outgoing bandwidth is limited.

Joomla Templates
Joomla Templates
Joomla Templates
Joomla Templates
Joomla Templates