Multi-Level Indexing in a Distributed Self-Organized Storage System

In many systems providing storage and retrieval operations on data, indices are used to make these operations more efficient. Distributed storage systems provide means to distribute the burden of storing and retrieving data onto multiple different computers. Routing indices can answer the central question in these systems: Where should one look for a specified data item? To be able to query for different columns in a relation or different entries in tuples, indexing for multiple dimensions is necessary. Our group applies a swarm-based approach to distributed storage leading to a new class of distributed systems, which are fully self-organized in their behavior and lack any shared global data structures. In this paper, we research whether multiple levels of routing indices can be maintained and used in such a distributed and self-organized storage service. To achieve this, we look into different types of indices and evaluate them in an experiment.