Extreme distributed systems

Ever since I joined the VU (back in 1995), I have been looking into large-scale distributed systems. For a number of years, my work concentrated on developing truly distributed objects in a system called Globe. When Guillaume Pierre joined the team, this focus shifted to large-scale collaborative Web sites (or actually: content delivery networks), resulting in the Globule system. Globule's unique feature is that it supports automatic replication of Web content at a fine level of granularity. Guillaume continued along that path and has expanded his view to Grid and Cloud computing.

My attention moved to further decentralization of distributed systems, which spawned an interest in epidemic-based  solutions. At this point in time I started to use the term extreme distributed systems as complete decentralization allows us to develop extremely large systems. Since about 3-4 years, I am also considering spatial distributed systems, notably wireless systems with extremely small nodes (cf. sensor networks). By-and-large, as long as the systems are extremely big, they have my interest.

Important: The people I collaborate with can be classified as experimental computer scientists. We validate our ideas through prototypes. Our systems act as instruments to understand, model, and validate observations.