High Availability and Performance of Database in the Cloud - Traditional Master-slave Replication versus Modern Cluster-based Solutions

Abstract

High availability (HA) of database is critical for the high availability of cloud-based applications and services. Master-slave replication has been traditionally used since long time as a solution for this. Since master-slave replication uses either asynchronous or semi-synchronous replication, the technique suffers from severe problem of data inconsistency when master crashes during a transaction. Modern cluster-based solutions address this through multi-master synchronous replication. These two HA database solutions have been investigated and compared both qualitatively and quantitatively. They are evaluated based on availability and performance through implementation using the most recent version of MariaDB server, which supports both the traditional master-slave replication, and cluster based replication via Galera cluster. The evaluation framework and methodology used in this paper would be useful for comparing and analyzing performance of different high availability database systems and solutions, and which in turn would be helpful in picking an appropriate HA database solution for a given application.