Authors:
Gergely Mezei
;
László Deák
;
Krisztián Fekete
and
Tamás Vajk
Affiliation:
Budapest University of Technology and Economics, Hungary
Keyword(s):
Graph Partitioning, Model Transformation, Cloud Computing, KL Algorithm, Social Networks.
Related
Ontology
Subjects/Areas/Topics:
MetaModeling
;
Model Tools
;
Model Transformation
;
Models
;
Paradigm Trends
;
Software Engineering
Abstract:
Dealing with extra-large models in software modeling is getting more and more common. In these cases, both memory and computational capacity of a single computer might be insufficient. A solution to overcome this barrier is to use cloud computing. However, existing algorithms have to be extended/modified to support cloud computing and use the advantages of its architecture efficiently. We focus on creating an algorithm to partition graphs representing models. Based on the algorithm, models should be able to be mapped onto several computational instances and processed in a distributed fashion efficiently. Previously, we have presented an algorithm that was based on the heuristic Kernighan-Lin partitioning method with two extensions: no limit on the number of partitions and not building on the knowledge of the whole model at beginning (nodes are received and processed one by one). However, when applying social network-based case studies, we have identified weaknesses of the algorithm.
This paper elaborates an enhanced algorithm that produces better results for extra-large models. Detailed measurements are also presented in order to show the improvement.
(More)