DBalancer: A tool for dynamic changing of workers number in storm

The workers´ number of a specific topology is fixed once it starts in Storm. This causes unnecessary waste of resources when workload shrinks down. In this paper, we propose an approach called DBalancer to dynamically increase or decrease the number of workers according to statistics of real-time data rate. DBalancer consists of three components: (1) Collector is used to collect data information about whether or not there is data at each period, (2) Decision Maker is used to statistic data rate information and make decision to increase or decrease the number of workers, (3) Rebalancer is used to execute decision made by Decision Maker. By applying DBalancer, we observe obvious lower memory occupation when workload comes down.