Would not some variants of Sybase be better, as the architecture is the same, and it has supported far more platforms which allow it to scale in ways that Microsoft SQL Server cannot?

I’d been running couple TB mixed olap + oltp on 2016 a few years ago and a single cluster handled what otherwise today would be 3 or 4 different data stores today. Performance of properly tuned column stores was literally mind blowing. Haven’t even needed memory optimized tables.

Can’t comment on Sybase. Didn’t need any other platforms.

Sybase and SQL Server came from the same code until release 4.8, so it would be familiar to you.

That was 30 years ago, though. Things have probably diverged quite a bit.

There were significant under-the-hood rewrites for SQL Server 2005 at least, and no doubt many smaller changes throughout the versions even where not adding new features, though there is still plenty that remains the same or at least very similar. I can name one bug from way-back-then which will never be fixed because maybe something somewhere “relies” on the odd behaviour!

I really gotta appreciate that someone's willing to say this out loud, rather than just pretending than MSSQL must be horrible because it comes from MS. If MSSQL was OSS Postgres wouldn't stand a chance.

This is ridiculous, Postgres is hands down a superior database at any price point.

Sorry, no. Not even close. Postgres is very good at what it can do and very cheap, but it just doesn’t do very much compared to SQL server or others. I’ve ran both mssql and Postgres in prod in the same project in perhaps not big but still nontrivial scale (low single digit terabytes) and SQL server made some things possible which Postgres simply didn’t have an answer for - column stores being the notable single killer feature if I had to name only one.

I'm currently running SQL Server and Postgres in the same project, and I would love to dump SQL Server. Its geospatial "capabilities" are so laughably bad, they should just stop advertising that they exist at all, and the unpredictable locking behavior alone should disqualify it from being used in any modern software project. And sure, it has column stores, which postgres has in extensions, but even if I needed column stores, SQL Server would still be last on my list, because there are so many better options out there.