Room: Baltic II+III
One of the biggest challenges regarding database upgrades is how to reduce the required downtime of a system when the upgrade is happening. It is not always possible to have a maintenance window for upgrade operations due to the unwanted effects of system outages such as financial costs, revenue loss, damaging the business reputation, the risk of not meeting Service Level Agreements (SLAs). The impact of the downtime may vary considerably across industries, also depend on the business size. In modern systems, zero- downtime upgrades is a big need and in fact, can be achieved through the advancements in IT automation, configuration management, and orchestration methodologies. Additionally, the emergence of Cloud Computing technologies enables applying software solutions at a scale, in cost and time-efficient ways.
The talk will cover different ways of upgrading PostgreSQL and will propose the automated tool to overcome downtime requirement. The tool utilizes pglogical (logical replication extension), pgbouncer (Postgres protocol aware proxy) and automates the processes using Ansible configuration management and orchestration tool. Developed tool is demonstrated on the servers in cloud, hence the talk will mention provisioning process of the Postgres cluster as well. Using the automated upgrade tool two case studies are performed. Benchmark results will be shared and the conclusion will include comparison with the existing upgrade methods as well.
This talk is mainly for people who are interested in DevOps and system engineering of PostgreSQL environments.
The following slides have been made available for this session: