2021-12-16 Solving PostgreSQL wicked problems

Description

Slide deck (PDF, 40 pages): https://github.com/orioledb/orioledb/blob/main/orioledb-postgresbuild-2021.pdf

The works is 2-part:

  1. prepare @PostgreSQL for even better extensibility (extrend custom AM, more custom things, new hooks) in Postgres,

  2. implement new ideas as extensions.

Code is to be published very soon.

I think this is one of the most interesting development happening with @PostgreSQL right now, wish Alexander success – especially with (1).

If it happens, Postgres users may benefit a lot.

Postgres extensibility is in its DNA. It is amazing how far it can go.

Today’s #RuPostgres stream was with Alexander Korotkov, one of @PostgreSQL contributors with a strong focus on performance.

His new startup is solving fundamental problems: 4-byte XID, bloat/vacuum, index write amplification.

Recent talk: https://vimeo.com/649616139 (video, 20 min)

../../../_images/postgresql_future.png

Alexander Korotkov

CEO of OrioleDB with key responsibility for the development of the transactional database engine.

Alexander was awarded a PhD for research on PostgreSQL algorithms from MEPhI, the leading and most prestigious Russian research institution and home to six (6) Nobel laureates.

Alexander stayed on to became an associate professor at MEPhI from 2011 to 2016 when his research was dedicated to improving the algorithms, multicore systems optimization and advance statistics, as well as selectivity estimation on the PostgreSQL platform.

In 2015 Alexander founded and become the chief technical and development officer of Postgres Professional in Moscow.

Alexander built a dedicated and highly successful team that assisted in the building of Postgres Pro Enterprise (a proprietary PostgresSQL fork) which is widely regarded as a breakthrough development for the effective employment of PostgreSQL in large enterprise and Government.

He is the inventor behind OrioleDB who seeks to deliver the next generation engine for the PostgreSQL and other open source DBMSes with significantly improved speed at a fraction of the cost.