[zackelan]

That strikes me as being comparable only at the most superficial level. I started that SQL Server container and it's using 825mb of RAM (according to systemd-cgtop), just sitting there idle with no data in it. I started a Postgres 9.6 container also using their official Docker container and it's sitting idle at 58mb of RAM.

Insult to injury, if you're using Docker in a VM (as you must on a Mac) the instructions on that Docker Hub page say you need a minimum of 3.25GB RAM dedicated to the VM. That's quite a heavy lift if all you want is a database instance for local development.

And then sure, you could run SQL Server on Linux for your production instances, but you're well off the beaten path. You're going to have to roll your own infrastructure rather than relying on AWS RDS, Google Cloud SQL, Heroku, etc etc. And the moment you have a performance problem the first advice you'll hear seems likely to be "try it on Windows and see if it reproduces there".

[mst]

That 58Mb RAM number isn't necessarily a good thing - something postgres has regularly fought against adoption-wise is that the defaults exist to be sane on a shared machine, not provide maximum throughput on a dedicated database server. While I appreciate this choice personally, I also suspect a good container default would idle higher.

(I'm a huge pg fan if that isn't clear - just ...)

[manigandham]

SQL Server does use it's own mini-OS layer and this what lets it virtualize and run on Linux too, so there are higher system requirements.

It seems you're just guessing on that last part though - SQL Server has been in production for decades and already has lots of tooling and there are even new CLIs available for the Linux environment. Production SQL Server comes with support from MS, you're definitely not on your own nor will they just tell you to run it on Windows, this is simply not how enterprise support works and they would have a hard time staying in business with answers like that.

We can discuss support issues but it's best to leave the assumptions and hyperbole out of it.

[_jal]

"Assumptions and hyperbole"? That MSSQL isn't available as a service like RDS? That tooling and integration that's been building up for decades around PG (or Mysql/Maria for that matter) doesn't exist? These are just facts.

I'm sure you're right that MS would be happy to sell support. What I don't see is a reason to care, unless I run in to a Windows app that can't use another DB that I suddenly need to run under Linux.

[manigandham]

I'm not sure what you're arguing. MSSQL is available on RDS and to quote the parent comment:

> ...SQL Server on Linux for your production instances, but you're well off the beaten path...first advice you'll hear seems likely to be "try it on Windows and see if it reproduces there"...

You dont think this is hyperbole? Have you actually used SQL Server or Microsoft support before? They do not treat enterprise customers this way and both Linux and MSSQL have plenty of tooling available. There's nothing challenging here.

> What I don't see is a reason to care

You're not the target market then. MSSQL has some compelling features and a big ecosystem, adding cross-platform support and a much easier installation process only helps customers with more options for deployment.

[_jal]

Oops, you're right; I thought it wasn't on RDS. My mistake, and a good reminder to double check these things.