PostgreSQL is a fantastic general purpose tool for storing data, but often overlooked for tasks it is actually ideally suited. Bespoke databases seem to crop up daily in the name of performance or functionality. This talk will examine the field of "time series" databases and look in depth as to how PostgreSQL can be used for the purpose. Databases of this nature have seen an explosive resurgence in recent years, and are often employed in monitoring systems to collect system and application metrics, but also in the growing world of "IoT".
Relational databases have been storing time-series data for a long, long time, so why are they becoming increasingly unpopular for such a classical purpose? What seems to be happening is the convergence on a simplified data model and access pattern, leading to the emergence of more "out-of-the-box" solutions. PostgreSQL is more than capable of storing data of this nature, and at considerable scale.
This talk will walk through the design and performance analysis for an efficient time-series architecture using only PostgreSQL constructs. Primarily aimed at more novice users of PostgreSQL, but hopefully those more experienced will find something interesting in the mix too.
The following slides have been made available for this session: