One of the the keynotes addresses to PGConf NYC was a talk by a technology manager at Goldman Sachs who had a bunch of interesting insights into the ecosystem. In their environment, Postgres is one of a number of supported platforms, so they have a good position for comparison (including to a nameless “commercial database”, the name of which can probably be easily inferred).
The most interesting insight to me is that for large companies, the marginal cost of installing a database is very low: most of the cost of Oracle is having it at all, because you typically aren’t provisioning hardware, etc just to install a new application. On the other side, even if you remove the licensing costs, the total cost of ownership of an open-source database is still quite significant, given that someone still needs to build tooling around the database (e.g. for replication, auditing, training, monitoring, and the like). It’s also useful for a large company to have someone to call one way or the other, as you can’t post anything that might give away secrets in public lists.
Also of interest, there is a lot of talk at the conference around audit trails (and Auditors and Regulators) which dovetails nicely with the notion of immutable data. As this is a topic of interest in parts of the wider tech community, I’d expect a lot of developments in this area in the future.
He suggested a few new features that might be valuable: more programmatic configuration (vs. just editing files), a service name concept (similar to Oracle’s tnsnames, I assume) where you can address a database as a service without necessarily knowing where it is (similar to DNS), and one of my favorites, APIs that run alongside SQL (other ways to access the contents of the DB).
Overall I find it very insightful to see how different people are running their infrastructure, with varying scale and business constraints.