Relations with literature:
Based on papers dealing with applications based on XQS/MD*Crypt and further more based on publications on advanced topics of distributed computing a presentation on experiences and development paths of XQS/MD*Crypt as means of education and computation will be given.
Relevance for statistics and informatics:
Net based techniques have become important in statistic computing. The need to solve large scale data intensive problems geographically independent led to client-server-architectures. With the emergence of the Internet, client server architectures allow access to high performance computers from almost every point in the world.
With MD*Crypt we have developed a package that provides access to the statistic environment XploRe. Clients based on MD*Crypt technology are used in statistic education to give students the possibility to use high performance environments on high performance platforms. Furthermore students and developers can adjust clients since MD*Crypt uses an event model which provides freedom in handling the computational results. The paper shows how widespread accessibility is achieved.
As 'distributed computing' is not only defined as accessing single machines geographically independent but also as sharing resources to solve large scale problems I would like to focus on this topic also.
How to achieve 'distributed computing'. As a part of this paper I want to show how to obtain distributed computing based on the XploReQuantletServer(XQS)/MD*Crypt architecture. I want to show ways to apply techniques of peer2peer grids to XQS/MD*Crypt architectures in order to enable distribution of large scale data intensive problems among several XploReQuantletServers.
Main references:
avaliable on the web:
http://www.mdcrypt.com
http://www.xplore-stat.de/java/java.html
publications:
'Client/Server based Statistical Computing'; Kleinow, Lehmann;
Computational Statistics; forthcomming 2002
Keywords:
distributed computing, education, grid-computing, XQS, MD*Crypt
