Data migration rate of the CRUSH-based distributed object storage with dynamic topology

Distributed systems are widely used to solve problems that require large computational or storage resources. The scalability of such systems makes them cheaper. However, the overhead for maintaining the system’s performance or operation ability can be significant. This paper considers a distributed P2P storage system in uncontrolled dynamic environment. The change of the structure or the topology in such system can lead to data migration that may cause system overload. This paper examines the intensity and the amount of these migrations for different CRUSH-based data placement approaches.