TM8101 Dependability Analysis of Information and Communication
Systems
Norwegian title: Pålitelighetsanalyse
av informasjons- og kommunikasjonssystem
Möbius for Stochastic Petrinet/SAN modelling.
How to run Möbius on the Pride (Wolfpack)
cluster:
- From a windows PC (Thanks to
Karin who has been the "guinea pig")
- From Linux or Mac OS X. As usual you are expected to manage by
yourself :-) However, take a look at the first and last item in the
windows PC guide:
- Get an account
- Start the X server
- Log on pride/wolfpack and start Möbius.
We have a noncommersial licence. If you
insist on installing Möbius on your Linux or windows machine (see
here for supported
platforms) for such use, contact
me.
Note that the Möbius manual are
rather cursor with respect to introduction of the UltraSAN modelling
concepts and constructs. Hence, to familiarise with these, the students
should use the older UltraSAN manual available in the course's paper
repository.
Tools for Markovian reward models
- Qitao's presentation.
- Haverkort's 1996 paper available elsewhere.
Recommended background (Anbefalte forkunnskaper):
Knowledge of the design of
dependable/fault-tolerant systems. Fundamental knowledge of
dependability modelling and analysis, including the use of
discrete state continuous time Markov models. Previous knowledge
requirements are covered by the course TTM4120 Dependable systems.
Required background (Forkunnskapskrav):
None
Objectives (Læringsmål):
A good insight into the options, methods and challenges related to
modelling, analysis and dimensioning of information and communication
systems, with both HW, SW and network subsystems. Emphasis
is put on
dealing with conditions that are specific for (highly) dependable ICT
systems like rarely occurring events, decisions based on few
observations and handling of large unstructured state spaces.
Topics (Faglig innhold)
Modelling with state diagrams (discrete
state continuous time Markov chains) and stochastic Petrinets.
Modelling of systems with distribution, fault-tolerance and
co-operating SW objects. Pros and cons of methods with respect to
problems addressed. Analytical methods for state based systems,
included system times, the rate of (rarely occurring) system failures,
measurement of interval availability and availability guarantees,
truncation of state spaces. Simulation techniques to provoke rare
events like importance sampling and splitting.
Teaching (Læringsformer og aktiviteter):
Colloquia/interactive lectures, where
it is expected that the students have familiarised themselves with the
topic on beforehand. Optional exercises.
Syllabus (Kursmateriell)
Journal and conference papers, etc. Announced at the beginning of the
term.