PGConf.SPB 2023

Traditionally, fault tolerance in Postgres is implemented using built-in replication mechanisms and external utilities that monitor the state of running Postgres instances and react accordingly when various failures occur. In this presentation, I will tell you what we like and what we don’t like about this approach, which alternative we see, what we have been able to achieve to date and what we want to get done by the time of release, which is planned for December.

During the years of its existence, Postgres Pro piled up a pool of problems when query execution was inappropriately slow or a query was too expensive to be executed, so it was never executed. Almost always in our practice, this was due to the choice of a non-optimal query plan. In our story, we will talk about a very unconventional attempt to solve this problem by re-planning queries. We will tell you what it is, how it works, who will find it helpful and the prospects for using this feature.

In modern data analysis, machine learning models are used as often as databases. Such IT giants as Google and Amazon have already combined them. Microsoft and Yandex are not far behind. Isn't it time to implement a machine learning model in PostgreSQL? In the report you will hear about the basics of machine learning, its implementation in databases and an example of realization as postres extension.

Recently, one of the most common migration methods has become migration using the free ora2pg tool. However, many note the extremely low performance of this tool, especially when it comes to databases of 1 Terabyte and above. This presentation describes how to significantly speed up database migration using oracle_fdw alone or in combination with ora2pg.