PgConf.Russia 2019

I would like to share my experience in optimization queries in PostgreSQL of RMIS (Regional Medical Information System).

It's so good when database behaves predictable. When the performance is lacking, you just add CPU cores, terabytes of RAM and millions of IOPS, and everything becomes good again. But it's rather unpleasant, when server have plenty of free resources, while database is still running slow. And it's especially sad if stress testing detects no problems, while real life workload of the same volume makes your database hang.

In this talk I will consider bottlenecks of PostgreSQL, which we met in our practice, and which causes sad behavior described above. I'll also explain what can be done at user level in order to evade these bottlenecks, and what developers are planning to do in order to eliminate those bottlenecks. I'm also planning give some recipes of stress testing, which could have to evade surprises in production.

My talk will be about different cases of usage and setup of the standby server; examples how to setup standby linked to your archive( to make an opportunity to recreate standby from archive after primary crashing and promoting your old standby); Avito experience of usage of standby server for read-only queries: problems and solutions; monitoring of standby.

The dark age for PostgreSQL started at 2GIS after transitioning to the microservice architecture. Every team tried to cook database on their own — by installing instances, juggling versions, trying to code deployments with numerous tools or using manual operations. It was the right time to develop a “silver bullet” — a common set of tools to solve all the problems at once. We created our own cluster solution based on well-known PostgreSQL, repmgr, pgbouncer and Barman. Despite of the complexity of our final solution, we developed a repeatable flexible deployment to accelerate postgresql cluster deployment and management. Also we deployed the our own cluster to consolidate all databases. It helped to eliminate team efforts for database management and focus on their main goals. Failover works, we tried it :-)