Load balancing and resource efficiency in large scale distributed graph analytics

Type Coffee Hour Talk
Start Date April 17, 2017 03:00 PM
End Date April 17, 2017 04:00 PM
Location 5317 Sennott Square
Organizer Name
Speaker Name Kenrick Fernandes
Abstract A number of popular distributed systems for graph analytic adapt parallel programming models in which work can be shared easily across machines. However, these can lead to inefficient use of resources due to the irregular structure of large graph data and dynamic behavior of analytic algorithms. In these situations, static apriori graph partitioning is inadequate to achieve balanced distribution of work. In this short talk, we will look at the dynamic behaviors of distributed graph processing systems and propose solutions to reduce load imbalance and increase resource efficiency.