Browsing by Subject "Communication pattern"

Filter results by typing the first few letters
Now showing 1 - 2 of 2
  • Thumbnail Image
    ItemOpen Access
    Investigation of load balancing scalability in space plasma simulations
    (Springer, Berlin, Heidelberg, 2013) Türk, Ata; Demirci, Gündüz V.; Aykanat, Cevdet; Von Alfthan, S.; Honkonen I.
    In this study we report the load-balancing performance issues that are observed during the petascaling of a space plasma simulation code developed at the Finnish Meteorological Institute (FMI). The code models the communication pattern as a hypergraph, and partitions the computational grid using the parallel hypergraph partitioning scheme (PHG) of the Zoltan partitioning framework. The result of partitioning determines the distribution of grid cells to processors. It is observed that the initial partitioning and data distribution phases take a substantial percentage of the overall computation time. Alternative (graph-partitioning-based) schemes that provide better balance are investigated. Comparisons in terms of effect on running time and load-balancing quality are presented. Test results on Juelich BlueGene/P cluster are reported. © 2013 Springer-Verlag.
  • Thumbnail Image
    ItemOpen Access
    Network-aware virtual machine placement in cloud data centers with multiple traffic-intensive components
    (Elsevier BV, 2015) Ilkhechi, A. R.; Korpeoglu, I.; Ulusoy, Özgür
    Following a shift from computing as a purchasable product to computing as a deliverable service to consumers over the Internet, cloud computing has emerged as a novel paradigm with an unprecedented success in turning utility computing into a reality. Like any emerging technology, with its advent, it also brought new challenges to be addressed. This work studies network and traffic aware virtual machine (VM) placement in a special cloud computing scenario from a provider's perspective, where certain infrastructure components have a predisposition to be the endpoints of a large number of intensive flows whose other endpoints are VMs located in physical machines (PMs). In the scenarios of interest, the performance of any VM is strictly dependent on the infrastructure's ability to meet their intensive traffic demands. We first introduce and attempt to maximize the total value of a metric named "satisfaction" that reflects the performance of a VM when placed on a particular PM. The problem of finding a perfect assignment for a set of given VMs is NP-hard and there is no polynomial time algorithm that can yield optimal solutions for large problems. Therefore, we introduce several off-line heuristic-based algorithms that yield nearly optimal solutions given the communication pattern and flow demand profiles of subject VMs. With extensive simulation experiments we evaluate and compare the effectiveness of our proposed algorithms against each other and also against naïve approaches.