University of Cambridge > Talks.cam > Computer Laboratory Opera Group Seminars > Explorations in the Grid Computing Jungle

Explorations in the Grid Computing Jungle

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Pedro Brandao.

Slides available .

Note the unusual time

Abstract: Grid Computing is the new highly publicized distributed computing paradigm that aims at federating computing resources across large area networks to provide ubiquitous, on-demand computing power. It has recently evolved towards Web-Services, another very popular distributed computing technology that has been proposed as the means to integrate software over the Internet into large scale added-value chains of services. Grid Computing and Web Services are fairly recent technologies, with multiple facets (SOAP, WSDL , WSRF, etc.), involving numerous building blocks (XML, .Net, HTTP , web servers, J2EE , etc.). It is sometimes difficult to understand how all the different components relate to each other, and how mature these technologies are. In this talk we will report on our attempt to better understand the java core of the Globus platform, a reference implementation of the current grid computing standards. In particular, we will present our efforts in tracking down the causes for amazingly bad performance of the platform. As we had very little knowledge of Globus, which is a large and highly complex piece of software, we had to develop pragmatic approaches to speed up our understanding with appropriate reverse engineering techniques. In particular, we will present the dynamic layering technique we have used to infer the structure of Globus, and how we used this technique to visualize the raw profiling results we obtained and better understand how each component of the underlying Java platform contributed to the system’s behavior. This is an ongoing, very practical work, that we think offers interesting insights into modern middleware developments. It may also have us wonder how long known problems might mix with new ones, and to which extend current abstraction and reuse techniques may be useful for rapid software development.

Bio: François is interested in developing open and principled middleware solutions for complex systems such as grid and large scale sensor networks. He is particularly interested in problems of fault-tolerance and resilience, and how aspect-oriented programming and computational reflection can be applied to provide these properties in the above systems.

François has been a lecturer at Lancaster since January 2005, after an intervening spell as a post-doctoral researcher at AT&T Shannon Laboratory (NJ, USA ), on an INRIA scholarship. He received his PhD in January 2004 for his work at LAAS -CNRS (France) on multi-level reflection applied to fault-tolerant systems. At LAAS , he has worked among others on the EU ISP Dependable Systems of Systems project (IST-1999-11595), and the EU Cabernet Network of Excellence. At Lancaster he is co-investigator on the Divergent Grid project (EPSRC EP/C534891), and the EU FP7 Diva project (Dynamic Variability in complex, Adaptive systems).

More information on Lancaster’s web page and on François’s personal web page .

This talk is part of the Computer Laboratory Opera Group Seminars series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2019 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity