Overview of the Research Group
Welcome at the web pages of the Parallel and Distributed Systems research group (Arbeitsgruppe Parallele und Verteilte Systeme - PVS). The group has been established in October 2010 and is lead by Prof. Dr. Artur Andrzejak.
Our research is devoted to issues of efficiency in large-scale scientific computing and enterprise systems and targets the following problems:
- How to utilize low-cost and non-dedicated hardware resources without sacrificing availability and service guarantees?
- How to ensure dependable operation of complex software systems with low management effort?
- Which algorithms and frameworks facilitate efficient processing of large amounts of data?
- How to program and configure such frameworks more easily?
In search for solutions we combine approaches from performance evaluation, machine learning/statistics, algorithm design and others to model, evaluate and prototype algorithms, frameworks and software systems.
Current Research Projects
Our active research projects within the above-described scope include:
- Decision models and scheduling strategies in low-cost Cloud Computing environments (e.g. Amazon EC2 Spot Instances)
- Machine learning / data processing algorithms designed for both parallel and incremental operation
- Online and hybrid-architecture (i.e. shared-memory / distributed) Map-Reduce frameworks
- Detection of Software Aging phenomena and their elimination.
Researchers interested in our work and students who would like to participate in these projects (via bachelor/master thesis work or via "Softwarepraktikum") are encouraged to contact us for further details.