Adaptive Load Balancing based on Machine Learning for Iterative Parallel Applications

Anna Victoria C R Oikawa (1) , Vinicius Freitas (1) , Márcio C Castro (1) , Laércio Lima Pilla (2)

1 UFSC - Universidade Federal de Santa Catarina = Federal University of Santa Catarina [Florianópolis]
2 ParSys - LRI - Systèmes parallèles (LRI)

The performance of irregular scientific applications can be easily affected by an uneven distribution of work among the computing resources. In this context, Load Balancing (LB) stands as one of the most important solutions to improve resource utilization. However, choosing the best-performing load balancing algorithm for a given application is not a trivial task. For instance, manually and statically choosing an LB algorithm does not work in situations where applications have a dynamic or unknown behavior. In this context, we propose a Machine Learning-based Adaptive Load Balancer (ADAPTIVELB) to automate the load balancing algorithm decision at run time. This approach monitors and collects information about the application dynamically, and according to the analyzed data, it makes a decision of invoking the most suitable LB algorithm. Our experiments show that ADAPTIVELB can select a good load balancing algorithm in most of the cases, leading to performance improvements over statically chosen LB algorithms and over the absence of a load balancer.

Mots clés: dynamic load balancing; machine learning; performance

Domaines: Calcul parallèle, distribué et partagé [cs.DC]

Anna Victoria C R Oikawa, Vinicius Freitas, Márcio C Castro, Laércio Lima Pilla. Adaptive Load Balancing based on Machine Learning for Iterative Parallel Applications. 28th Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP), Mar 2020, Västerås, Sweden. ⟨10.1109/PDP50117.2020.00021⟩. ⟨hal-02570549⟩