Overall Satisfaction with PostgreSQL
We use Postgres for both OLAP and OLTP use cases. We use it as our data warehouse, for interactive queries, and for storing application data. Postgres is one of our main data warehouses, and we use it in congress with BigQuery to store, analyze and finally index data into our ElasticSearch cluster. One of our primary uses of Postgres is for geospatial analytics, so we leverage the PostGIS extension extensively.
- Spatial Analytics and other GIS use cases - PostGIS is an excellent way to get into spatial analytics, loading it up with data is trivial, power is on par with commercial solutions.
- Interactive queries over large (but not huge datasets) - easy to load data, query it with standard SQL, easy to set up and maintain.
- Support for a variety of data types - storing data in the database using semantic types is helpful for deeper analysis.
- Clustering -- we'd love to see clustering built into the product itself instead of third-party
- Parallelization -- PG is already going in this direction, but it will take a few more releases to be there
- Tooling -- we use a third-party tool right now to query PG, would love to see a first-party quality query tool
- Self-hosted without licensing fees means we are able to save thousands per year on licensing fees
- Ability to grow into a hosted environment with our database means that we're not stuck when we outgrow our ability to operate the system
- Immense support for third-party plugins means that when we have a use case that stock Postgres can't meet, we are likely to find an open source or commercial solution that plugs right into our DB
We evaluated MySQL, Postgres, MSSQL Server and a few other hosted options. We loved the fact that PG is both a GIS system and a traditional RDBMS. The mindshare around PG is high -- thousands of answers on StackOverflow, very active developer community and a punctual release schedule. PG is a standout open source product.