Try TimescaleDB: check how it would be helpful for partitioning of large tables, benchmark, collect cons and pros
As we discussed several times in the past, TimescaleDB, even without its "distributed" version (added in 2.0), is a popular tool that helps manage time-decay / time-series data, automates partition management, added powerful compression.
The goal here is to perform basic tests in the benchmarking environment and see if and how TimescaleDB can be useful for further scaling GitLab's databases.
Edited by Nikolay Samokhvalov