I'm sort of the opposite of GP. Having worked with SQL Server the first 5-6 years I've transitioned to using Postgres more over the last 3-4 years and really prefer it. That said SQL has some nice features, off the top of more head:
* It's pivot/unpivot query syntax is a lot nicer than the comparable Postgres extension.
* The ability to define a clustered index for a table that's maintained automatically.
* The ability to define materialized views that are maintained automatically.
* The db engine can automatically use materialized views to accelerate queries matching the view definition.
* It has some really cool support around temporal tables. Not just defining them, but using them in queries. Like, there's a pretty simple query syntax that lets you join together multiple temporal tables and say, show me what the data looked like at this point in time, or show me how it changed during this time period.
The main problem with SQL Server IMO is that many of the cool features you discover are gated behind a SQL Server Enterprise license, and it's crazy expensive. Upwards of $10k per core per year, IIRC. The last time I looked at RDS instance prices I believe SQL Server Standard was about 4x more expensive than Postgres, and Enterprise about 8x more expensive.
Edit: as a sysadmin the quite easy to configure default maintenance plans with backups is a godsend. It is still very awkward what you need reinvent the wheel (or use/buy a 3rd party product) to just backup databases on schedule in PG/My SQL.
* It's pivot/unpivot query syntax is a lot nicer than the comparable Postgres extension.
* The ability to define a clustered index for a table that's maintained automatically.
* The ability to define materialized views that are maintained automatically.
* The db engine can automatically use materialized views to accelerate queries matching the view definition.
* It has some really cool support around temporal tables. Not just defining them, but using them in queries. Like, there's a pretty simple query syntax that lets you join together multiple temporal tables and say, show me what the data looked like at this point in time, or show me how it changed during this time period.
The main problem with SQL Server IMO is that many of the cool features you discover are gated behind a SQL Server Enterprise license, and it's crazy expensive. Upwards of $10k per core per year, IIRC. The last time I looked at RDS instance prices I believe SQL Server Standard was about 4x more expensive than Postgres, and Enterprise about 8x more expensive.