Comparative Performance Analysis of PostgreSQL High Availability Database Clusters through Containment

Abstract: The enormous amount of data flow has made Relation Database Management System the most important and popular tools for persistence of data. While open-source RDBMS systems are not as widely used as proprietary systems like Oracle databases, but over the years, systems like PostgreSQL have gained massive popularity. High-availability database clusters (also known as HA clusters or failover clusters) are groups of computers that support server applications that can be reliably utilized with a minimum of down-time. This article is an attempt to set a benchmark of PostgreSQL high-availability databases in comparing the performance of same-containment keepalived-repmgr clusters against cross-containment HAProxy-PgBouncer clusters. The result shows that our cross-containment HAProxy-PgBouncer is still a significantly better performer with load balancing, healthcheck and its throughputs are improved from 0.346% to 9.454% in compared to keepalived-repmgr. Also we would like to present I/O activities and CPU usage percentages between the two kinds of PostgreSQL HA clusters.

Keywords: Altus cloud, cross containment, failover, HAProxy, healthcheck, high availability, keepalived, load balancing, PgBouncer, PostgreSQL, repmgr, repmgrd.