NMRG/EMANICS Workshop Position Statement

Aiko Pras <a.pras@utwente.nl>
University of Twente

In the next decade the main challenge in networking will be how to
improve dependability of the IT infrastructure, including the services
running on top of it. The issue will no longer be how to make under good
weather conditions networks a few percent faster, but how to ensure,
even in case of "bad weather" (DOS attacks, failing equipment) a      
predictable service on which users can rely.

Dependability is a goal, however, and not a kind of mechanism that can
easily be implemented. In this sense dependability is like security,
performance and scalability, which are also goals. The mechanisms
(protocols, algorithms etc.) needed to make an infrastructure dependable
will vary from case to case. It is essential, however, that
researchers start from real problems, as identified by real managers,
and not from artificial problems that look well on paper but have little
relationship with reality. Examples of real problems are: how to keep
viruses outside, how to fight spam, how to immediately identify system
components that do not perform well, how to find inconsistencies in my
configuration etc.