Towards making big data applications network-aware in edge-cloud systems

The amount of data collected in various IT systems has grown exponentially in the recent years. So the challenge rises how we can process those huge datasets with the fulfillment of strict time criteria and of effective resource consumption, usually posed by the service consumers. This problem is not yet resolved with the appearance of edge computing as wide-area networking and all its well-known issues come into play and affect the performance of the applications scheduled in a hybrid edge-cloud infrastructure. In this paper, we present the steps we made towards network-aware big data task scheduling over such distributed systems. We propose different resource orchestration algorithms for two potential challenges we identify related to network resources of a geographically distributed topology: decreasing end-to-end latency and effectively allocating network bandwidth. The heuristic algorithms we propose provide better big data application performance compared to the default methods. We implement our solutions in our simulation environment and show the improved quality of big data applications.