PGConf.Russia 2018

Patroni is a Python application to create high-availability PostgreSQL clusters based on the streaming replication. It is used by Red Hat, IBM Compose, Zalando and many other companies. This tutorial will highlight Patroni architecture, provide attendees with hands-on experience of configuring high-availability PostgreSQL clusters with Patroni, describe how to take advantage of numerous additional features and give an opportunity to learn more about common mistakes related to running Patroni and its troubleshooting.

In order to take most out of the Patroni tutorial one needs a laptop with git, vagrant and virtual box installed.

Vagrant can be obtained from https://www.vagrantup.com Virtualbox is at https://www.vagrantup.com

Alternatively, one can install your Linux distribution packages (or use homebrew on Mac).

Once Vagrant and Virtualbox are installed one can run the Patroni VM by issuing the following commands:

$ git clone https://github.com/alexeyklyukin/patroni-training
$ cd patroni-training
$ vagrant up

When the setup concludes Patroni box can be accessed via ssh using vagrant ssh command.

How to optimize query processing in PostgreSQL? What if we deal with hundreds of servers and thousands of instances? The company "Tensor" has developed a special tool - explain.sbis.ru, which enables synchronous collection and analysis of queries, visualization of implementation plans, and monitoring of bugs in database.

Lately, the PostgreSQL is more widely being used for Enterprise. Our company, Ingram Micro Cloud, is one of the first companies that did it. We have been using PostgreSQL for many years as the main DBMS for our products. In our report, we want to tell about the evolution of our High Availability (HA) PostgreSQL cluster. We will tell about how quickly we implemented the solution using pgpool-II, wrote failover scripts, tested Postgres-XL, and came up with unusual configurations of Stolon. We will also cover the problems of load balancing, pooling connections, and backups.

WAL-G is simple and effective disaster recovery tool for PostgreSQL using cloud storages. In its core functionality, WAL-G is the successor of WAL-E rewritten in Go. But there is one new neat feature - delate-backups. WAL-G delta-backups, whenever possible, stores only pages, changed since the previous backup. In this talk, I'm going to describe development process of this feature.

Surprisingly, most important and complicated question was the design of the interface: WAL-e is simple and comprehensive, keeping these properties was goal #1. Technical details of implementation were covering some underwater stones too. Besides these, I want to discuss the perspective of technological development and future coordination of recovery tools developers.