PGConf.Russia 2023

With the rise of NoSQL databases, a number of falsehoods have flourished regarding how to choose a database engine. This talk focuses specifically on Redis and PostgreSQL, and why one might choose one or the other.

At small scales, we can often get by thinking of database servers as black boxes, but as we scale, the internals and architecture become more and more important. This talk focuses on behavior of these systems at scale and under load.

In this presentation you will learn:

• How Redis and PostgreSQL differ architecturally
• How differences in architecture affect scalability and performance
• Cases here Redis is the clear winner
• Cases where PostgreSQL is the clear winner

Additionally, some notes will be offered in terms of where PostgreSQL can improve in to compete with the sorts of workloads that generally favor Redis.

The following topics will be covered as part of the presentation:

• Architecture of logical replication
• Publisher introduction
• Subscriber introduction
• Data syncronization introduction
• Logical decoding
• Replication slot
• output plugin

My presentation "Explaining the Postgres Query Optimizer" covers the details of query optimization, optimizer statistics, joins, and indexes. This talk covers 40 other operations the optimizer can choose to handle complex queries, large data sets, and to enhance performance. These include merge append, gather, memoize, and hash aggregate. It explains their purpose and shows queries that can generate these operations.

This is a new talk; draft slides are at https://momjian.us/main/writings/pgsql/beyond.pdf

Доклад о нашем опыте внедрения PostgreSQL вместо Oralce в высоконагруженную систему с микросервисной архитектурой, обрабатывающую несколько террабайт данных. Расскажем о том, как прошли путь от пилотирования Greenplum до перехода на несколько взаимодействующих баз PosgreSQL с разными профилями нагрузки, и о проблемах, с которыми столкнулись в процессе.