The Peer to Peer research project investigates innovative decentralised techniques to manage networks and data to provide complex applications efficiently and economically.
Service providers have been using the traditional client-server approach to provide online applications such as webcasting, video-on-demand and network gaming, often to millions of internet users. This involves the use of one or more dedicated servers deployed at strategic locations in the network, which users connect to and request services. However, this approach has several drawbacks:
- Server capacity needs to be upgraded as the demand for a service increases
- Rich media content requires the underlying network to be upgraded to handle the increase in network traffic
- Central servers are a single point of failure and require constant maintenance.
These problems have forced telecommunications and service providers to explore alternative methods for managing the network in an effort to provide these services efficiently.
What will this research achieve?
The aim of this project is to explore the peer-to-peer paradigm as an alternative technique for network management and content delivery. It will attempt to solve the problems of the client-server approach so that service providers can offer online services efficiently and economically.
Who will benefit?
Internet service providers will be able to deliver services more cost effectively and users will be able to access a wider variety of services. The peer-to-peer model will also help scientists access large-scale computing resources at a low cost.
What are the key features?
We are currently investigating the peer-to-peer networking model in a variety of contexts:
- Massively multiplayer online gaming network engine. A decentralised platform to develop and deploy online games and virtual worlds efficiently and cost effectively. The technology would reduce the cost of maintaining expensive game servers by delegating data processing to individual participants. This would also improve resilience to failures by removing the single point of failure. It would reduce game traffic in the core network, improving the system performance.
- MOLAN. A platform for developing sophisticated applications for handheld devices in a wireless environment. The exchange of data for specific applications on a mobile ad-hoc network requires a new class of algorithms because some of the assumptions used for ad-hoc mobile networks do not hold for overlaying sophisticated applications. We are developing a decentralised solution for high-performance computing using the peer-to-peer paradigm which eliminates the need for high-performance computing centres, reducing hardware costs significantly.
- eScience Desktop Peer. This system uses the peer-to-peer model to allow scientists to share resources over the internet, delivering high performance with a low maintenance cost. The decentralised system architecture ensures the system is scalable and easy to maintain. It allows scalable checkpoint image storage and recovery, which is a bottleneck for a centralised architecture, and provides a viable basis for building grid applications.
The project team has developed Badumna network engine.
Badumna is an advanced network engine designed for virtual worlds and massively multiplayer online games. The highly scalable and cost effective platform that utilises a peer-to-peer network can support millions of users with very minimal infrastructure. Badumna suite comes with a well defined set of APIs that provide multi-user functionality, be it for 10 users or 10 million.
Badumna provides a massive competitive advantage over traditional network engines as it reduces the hosting costs by more than 80%. Using the Badumna platform, virtual worlds developers can create large-scale, highly scalable virtual worlds that are interlinked to each other.
For more information about Badumna network engine, please visit: www.badumna.com
Research team
Santosh Kulkarni (project leader)
Scott Douglas
Aaron Harwood (15%)
Egemen Tanin (15%)
Lars Kulik (10%)
Shanika Karunasekera (15%)
David Churchill
Further Information
For further information contact Santosh Kulkarni: santosh dot kulkarni at nicta.com.au or click here.