Re: Some Obvious Reasons..... @RICHTO
On ACLs.
Proprietary UNIX has had filesystem ACLs of the type you are talking about since at least 1990. I am most familiar with AIX, and this was a major enhancement when the RISC System/6000 was launched in 1990 with AIX 3.1.
The Posix 1 filesystem permissions were a description of the original UNIX permissions model that was invented back in the 1970s before Microsoft even existed. At that time, the most sophisticated security model around was that proposed for Multics, many features of which made it into both VMS and PrimeOS (and it is worth remembering that Richard Cutler had some responsibility for VMS).
This is for a filesystem, I admit, but the basis of Role Based Accounting (acquired credentials used to control running processes and services) was introduced in AIX in 4.3.3, which IIRC was around 1998.
If you look outside of core UNIX, then DCE/DFS, which was a standards based enhancement which sat above the OS, and worked on various UNIX OS's, OS/2 and even windows NT provided ACLs for processes and file objects around 1994, and this was based on the Andrew File System (AFS) and Apollo's NCS which were earlier still. AFS, and DCE/DFS allowed credential management using Kerberos a long time before that support was integrated into Windows, and was provided by the OS vendors in most cases. AIX could build in a Kerberos based user authentication system from about AIX 4.2 in 1995.
I'm fairly sure that those people who were familiar with Veritas will also have something to say.
In terms of NFSv4, the Linux support may be experimental (which probably reflects more on the people doing the work than NFSv4 itself), but has been part of the core facilities provided by at least Solaris and AIX for quite some time (have to look up when it was introduced, but I remember reading up on in in 2005). Definitely not experimental on those platforms.
Having got that off my chest, it is clear that these arguments are pointless. This is because although I have a good knowledge of AIX and traditional UNIX, my knowledge of Windows is incomplete, so I so not make direct comparisons of capabilities. I suspect that there are actually very few people who are able to make a dispassionate comparison of these features between OSs, so just having a willy waving competition in forums such as this one is largely pointless.
That said, I do like the idea of a Windows Server that allows you to strip down the basic install to the minimum necessary to run an application. Seems consistent with KISS, one of the primary requirements to make any service functional and secure.
It is pointless to have more features than you need which may open up security or performance issues running on a server which has a specific defined function. This is where heavily (de-)configured Linux distributions have had a real advantage in the server space for years, because you could strip them down relatively easily to the bare minimum. It looks like Microsoft have finally learned.