Efficient Fault-Tolerant Transactions for Distributed Graph Database

  • Ray Neiheiser UFSC
  • Roland Schmitz Stuttgart Media University
  • Luciana Rech UFSC
  • Manfredo Manfredini The University of Auckland
Palavras-chave: Graph Databases, Distributed Systems, Fault Tolerance

Resumo

Through the ongoing trend in graph technologies due to the massive growth of linked data produced by social networks graph databases gained popularity.
Replication, a common approach to increase availability in databases, is also used by diverse graph database solutions. Few approaches implementing fault-tolerance in graph databases have been proposed yet.
This paper considers deferred update replication using atomic broadcast in order to implement fault-tolerance in distributed graph databases. The main contribution of this paper is a deferred update algorithm adapted to graph databases offering a more scalable and faster solution, showing a performance advantage of over 30\% compared to existing approaches.

Publicado
2019-10-19