Copenhagen Programming Language Seminar
Increasingly faster computer for both private and corporate use combined with the dissemination of faster network connections constitutes a still larger unused resource for storage and computation. This present new opportunity for large scale distributed computing.
By making a new combination of technologies that has not previously been used for this purpose we designed a system that seeks to harness these unused resources. The system is based on distributed hash tables and uses principles as ’deterministic key hashing’, ’pseudo-random task distribution’ and ’light/heavy state separation’ to achieve both scalability and fault tolerance. It is able to solve a broad spectre of problems, even problems with mutually depended subtasks. To the best of our knowledge this combination of properties makes the system stand unrivalled. The system is implemented in a prototype called Dipeon that has demonstrated to scale well and to solve different algorithms; combinatorial problems, branch & bound algorithms etc.
The presentation will be based on our master thesis from March 2003 and will cover the use of distributed hash tables both regarding routing and task distribution and deterministic location of storage for results. In addition we will discuss the principles for categorising data, the pros and cons of the design and which types of algorithms it is best suited for.
The presentation will be held in Danish
Peter Sestpft. Administrative host: Camilla Jensen. All are welcome.
The Copenhagen Programming Language Seminar (COPLAS) is a collaboration between DIKU, ITU and KVL.
To receive information about COPLAS talks by email, send a message to firstname.lastname@example.org with the word 'subscribe' as subject or in the body.
For more information about COPLAS, see http://www.coplas.org