FUGU: Elastic Data Stream Processing with Latency Constraints

Elasticity describes the ability of any distributed system to scale to a varying number of hosts in response to workload changes. It has become a mandatory architectural property for state of the art cloud-based data stream processing systems, as it allows treatment of unexpected load peaks and cost-efficient execution at the same time. Although such systems scale automatically, the user still needs to set configuration parameters of a scaling policy. This configuration is cumbersome and error-prone. In this paper we propose an approach that tries to remove this burden from the user. We present our data stream processing system FUGU, which optimizes the selected scaling policy automatically using an online parameter optimization approach. In addition, we demonstrate how our system considers user-defined end to end latency constraints during the scaling process.