PROJECT TITLE :

Adaptive Path Isolation for Elephant and Mice Flows by Exploiting Path Diversity in Datacenters

ABSTRACT:

Resource competition and conflicts in datacenter networks (DCNs) are frequent and intense. They become inevitable when mixing elephant and mice flows on shared transmission paths, ensuing in arbitration between throughput and latency and performance degradation. We tend to propose a unique flow scheduling theme, Freeway, that leverages on path diversity within the DCN topology to ensure, simultaneously, mice flow completion at intervals deadline and high network utilization. Freeway adaptively partitions the offered paths into low latency and high throughput paths and provides different transmission services for each class. A M/G/1-based mostly model is developed to theoretically obtain the highest price of average delay over the trail that will guarantee for ninety ninep.c of mice flows their completion time before the deadline. Based mostly on this certain, Freeway proposes a dynamic path partitioning algorithm to adjust dynamically with varying traffic load the amount of low latency and high throughput methods. While mice flows are transmitted over low latency methods using a simple equal price multiple path (ECMP) scheduling, Freeway load balances elephant flows on totally different high-throughput ways. We evaluate Freeway in an exceedingly series of simulation on a massive scale topology and use real traces. Our evaluation results show that Freeway considerably reduces the mice flows completion time within deadlines, while achieving remarkable throughput compared with current schemes. It's exceptional that Freeway will not would like any change of DCN switch materials or scheduling algorithms and can be deployed easily on any generic datacenter network with switches implementing VLANs and trunking.