ABSTRACT
This article addresses the problem of Data Replica Placement, an important technique used in storage-capable distributed networks to improve system availability, reliability, and fault-tolerance. The study focuses on the Data Replica Placement problem in strategic settings inspired by practical market-based data replication applications, such as content delivery networks. Multiple self-interested players with private preferences own data objects for replication. Players compete for storage space among replication servers for placing replicas with the objective to optimize their own profits. Using mechanism design approach, the authors consider the problem as a sequential composition of knapsack auctions and design an algorithmic mechanism DRPMECH to aggregate players’ preferences and approximate a social efficient allocation for the problem. This work analyzes both the economic and computational properties of DRPMECH, validates the properties using experiments, and compares its performance against related game-theoretical solutions.