PGConf.Siberia 2017

The highly available configurations for PostgreSQL are usually mode with the help of binary replication and some third party utilities. This talk will review the existing solutions, including a simple one based on corosync/pacemaker, which is supported by Postgres Pro. A live demo will be presented.

Another hight availability approach is proposed by the multimaster Postgres Pro extension, which provides high availability without any third-party products. A live demo and description of multimaster caveats will be presented.

Several years ago it was like that, you say “.Net developer” mean Windows and Microsoft SQL Server developer. But time flies, things change. Microsoft became a platinum member of the Linux Foundation, real .Net Core applications work on Linux/Unix servers, PostgreSQL's popularity goes up including .Net projects.

In my speech I will compare several popular ORM frameworks and tell how to make them work faster. I’ll tell about Dapper-Dot-Net “micro ORM” in a nutshell, that doesn’t generate SQL commands, it just maps DB-records to C# objects, but does it very fast. Also, it’s possible to combine different ORM frameworks with each other to improve performance.

The most interesting thing I want to talk about is how to accelerate insert and update operations using the peculiar properties of «Multi Insert» and «Upsert» in PostgreSQL. I’m gonna make a real-time demo of inserting and then updating one million records to PostgreSQL instance running on ordinary notebook with Windows in a just a few seconds, using ORM.

Table Partitioning is a highly demanded functionality of PostgreSQL. The existing possibility to make partitioning through inheritance has several serious disadvantages, such as ineffective planning, missing runtime optimization, need to manage triggers and sections manually.

This talk describes the pg_pathman extension developed by our team. pg_pathman supports HASH and RANGE partitioning and performs query optimization during planning and execution stages, allows fast includes by using custom executor nodes instead of triggers, supports FDW, lock-free data migration and other possibilities.

In conclusion I will describe the new features, expected in next releases, and our plans for integration with PostgreSQL 10.

The talk describes the existing solutions for backup and restore for PostgreSQL, including barman, pgbackrest and pg_probackup utilities. The comparison of their functions with analysis of theis drawbacks and advantages will be presented. You will learn how to setup a block level incremental backup with compression and integrity verification.