Dynamic service placement and redundancy: Concept and research direction

Service placement algorithms have been introduced to automatically manage services in distributed environments. However, most of these algorithms aim to improve service performance or to reduce operation cost and do not consider the reliability and availability of their resources. Most of them do not consider events when machines in their resource pool malfunction. These malfunctioned machines could cause services to halt or, worse, to produce transient and intermittent errors. In this paper, we introduce a dynamic service placement and redundancy (DSPR) concept to manage service placement and replicas to achieve high availability and that is capable of detecting and recovering from arbitrary faults. The DSPR algorithm will autonomously look for opportunities to ensure service availability by migrating services to hosts with better performance and availability. The concept, research methodology and potential evaluation techniques are included.