Efficient Model Partitioning for Distributed Model Transformations - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2016

Efficient Model Partitioning for Distributed Model Transformations


As the models that need to be handled in model-driven engineering grow in scale, scalable algorithms for model transformation (MT) are becoming necessary. Programming models such as MapReduce or Pregel may simplify the development of distributed model transformations. However, because of the dense inter-connectivity of models and the complexity of transformation logics, scalability in distributed model processing is challenging. In this paper, we adapt existing formalization of uniform graph partitioning to the case of distributed MTs by means of binary linear programming. Moreover, we propose a data distribution algorithm for declarative model transformation based on static analysis of relational transformation rules. We first extract footprints from transformation rules. Then we propose a fast data distribution algorithm, driven by the extracted footprints, and based on recent results on balanced partitioning of streaming graphs. To validate our approach, we apply it to an existing distributed MT engine for the ATL language, built on top of MapReduce. We implement our heuristic as a custom split algorithm for ATL on MapReduce and we evaluate its impact on remote access to the underlying backend.
Fichier principal
Vignette du fichier
sle2016.pdf (548.29 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-01367572 , version 1 (16-09-2016)


  • HAL Id : hal-01367572 , version 1


Amine Benelallam, Massimo Tisi, Jesús Sánchez Cuadrado, Juan de Lara, Jordi Cabot. Efficient Model Partitioning for Distributed Model Transformations. Proceedings of the 2016 International Conference of Software Language Engineering, Oct 2016, Amsterdam, Netherlands. ⟨hal-01367572⟩
201 View
330 Download


Gmail Facebook X LinkedIn More