Why would they do that? Linux desktop has no penetration into the market. It would be a big investment to capture an additional 1% market share.
I can see MS porting all their server apps(MSSQL, Exchange, etc) to run on linux since Linux is dominating the server market and it would be a significant opportunity for them.
IMHO thing an MS Linux could really add for SME IT teams is.
1) An easy way to join a Windows domain. Being able to use AD user auth and Windows file sharing with AD groups that is as easy as it is on Windows server would remove a major scare factor for Windows admins trying Linux. I hope this could be implemented by contributing to SAMBA and being a wrapper for SAMBA rather than a reimplementation. I have had frustrating times with Linux unmounting Windows shares and having to reboot to get them to remount.
2) Training. MS have a well trodden path for learning to be an MS admin. Bringing it to Linux would surely help more people test-the-waters.
Red Hat's sssd already makes this trivial. You can join a Windows domain from the Fedora "first run wizard" or via a simple "realm join xyz.tld" command. It works pretty well mostly. You get AD integrated logins and SSH/sudo access and can use AD groups for access control just like normal groups. Kerberos SSO to Linux boxes works fine etc.
AD is all about group policy, logins are merely table stakes. If you want to distribute firewall rules to every machine in your arg AD is the way to go. Ditto for forcing updates (or preventing them), and any other manner of mass configuration. Then there is also centralized logging and auditing.
Linux has most of this in an ad-hoc fashion, but the key benefit comes from the centralization.
SSSD makes joining AD easy for any linux. Suse has integrated it into Yast and it's super easy. There are also tons of linux training initiatives.
The simple fact is alot of Windows admins are incurious or set in their ways. It's true for Linux as well, but not as true.
It works quite well too. As I understand it, SQLServer itself is effectively running on top of its own "OS" like layer (even on windows), so porting cleanly was made significantly easier.
Seconded. SQL server does work quite well on Linux and was surprisingly easy to get going. I prefer Postgres but sometimes we don't always get what what we want and hey, I'll take it cause I can use it on Linux!
As for MS Windows, WTF? Not only the forced ads making it seem like in some kind of casino built for 4 year olds (assume it's some kind of play to suck the kiddies in?), it's an abortion of mismatched UI, half of it trying to look like a cheaply built web app, then on next screen dumping you back into some NT style select panel. It's just awful. To say nothing of invasive updates and general Windows unpleasantness. I can say without reservation I have no use for the OS at all, Linux is just so much more pleasant to use at this point.
Is there any advantage to running it on Linux (except maybe the cost of a Windows license which is probably trivial compared to the cost of Sql Server itself)
I haven't seen benchmarks but I would guess that there isn't going to be much of a performance difference. It's like comparing the performance of your VM on different hypervisors, unless something is very wrong you won't see much difference.
As the OP indicated, MS SQL has it's own facilities for managing memory and disk I/O. It basically asks for all the resources from the OS and then does everything itself internally. I'm exaggerating obviously but it does try to avoid costly calls to the OS whenever and however possible which is why it was so easily ported to Linux.
The Docker images are smaller for Linux for one. Our team use the Linux version in tests during builds just for that. We are relying that Microsoft manage to keep the two versions close enough that it's meaningful.
Azure is a cash cow. Windows brings in money but is increasingly difficult to monetize _and_ its increasingly difficult to keep bringing legacy app support forward and stay competitive.
PowerShell, .NET Core, and MS SQL are all happily running on Linux.
By all accounts WSL entered the market with little drama is pretty great to use (IMHO).
The idea of doing a native SSH from WSL into a Linux container on Azure running a bootstraped 'MS' version of *nix with official support for a discounted price compared to a Windows VM is certainly appealing.
I would run MS Linux as a dev environment in a heart beat. MS would do well to _not_ to brand a desktop and just run a very clean gnome or plasma default.
Then a dedicated app store can really push PWAs like Teams,Skype and fully manage the interop layer for other apps that need a little more hand holding (wine, mono, etc).
Given the tight cooperation between Canonical and MS for WSL I would somehow expect that MS would use Ubuntu if they were to ever sanction a desktop Linux. This is just my guess, though.
Windows is more or less over and MS has been gradually coming to terms with it, helped along by inflection points that made it painstakingly clear that the Windows way was beyond outmoded, like containers and cloud. I don't think MS is even pretending otherwise anymore.
I wouldn't be surprised at all to see MS give up on WSL, acquire CodeWeavers, and reimplement "Windows" as a proprietary desktop environment for a nix-ish OS with a super-souped-up WINE doing much of the legwork. At this point such a contraption would be less painful than some of Apple's recent transitions (e.g. from PPC to x86).
Selling software, as a general business model, is on the ropes and this is a great indication of that. The victory of open-source here is both blatant and decisive.
Had it not been for every major software company deciding they can subsist on a combination of a) rental fees and b) advertising/demographic data, we'd probably have another RIAA v. The Internet-style showdown to confront over the next decade. In this respect, I suppose we should be grateful for the opportunity to pay 6x more to be in "the cloud".
As an observer, it's a weird situation to see, and still trying to orient my feelings and understand what to make of it.
I would be very surprised to see them take that reimplementation path as their backwards compatibility has been so important for decades (yes I know not everything runs still; but a large amount does). They have provided an evolutionary path and that would be revolution. The benefit to them seems likely to be outweighed by the disruption. And they know points of disruption are where they are more likely to lose customers.
> At this point such a contraption would be less painful than some of Apple's recent transitions (e.g. from PPC to x86).
I find that hard to believe. Apple only had to build emulators for their ISA changes, just translating instruction set into another. Doesn't sound painful at all. WINE on the other hand has good reasons for pointing out that it's not an emulator.
WINE runs most applications extremely well. With actual backing from a MegaCo, especially the MegaCo that owns all of the IP around Windows, I have no doubt that a year of work would round out the rough edges such that compatibility differences don't exceed what would be expected between major versions of Microsoft Windows.
I'm not enthusiastic about this possible future, but here's why:
it means they could eventually just adopt Linux as the platform for Microsoft applications. That would do away with a lot of historical reliability issues, and would mean that they would have less to support internally. They're making more and more money on things like Azure and Office365 etc, and none of that really depends on Windows.
I don't think anything has changed and it definitely wouldn't make sense now. There is zero growth in mobile phones, in the US and globally overall. That isn't going to get better, it's already a stagnate, saturated market. Now would be about the worst time for Microsoft to decide to re-target that market.
Azure can be a ~$30 billion sales business eventually, probably spitting off $6b or $7b in operating income. It's a dramatically better business than messing with smartphones again (other than building apps for Android & iOS). They should just keep focused on their booming cloud business.
I don't think any of this is likely, even remotely.... but if it was, they wouldn't be motivated by adding that 1% to their user base. The motivation would be to deprive the competition of that 1% user-base: weaker competition.
I can see MS porting all their server apps(MSSQL, Exchange, etc) to run on linux since Linux is dominating the server market and it would be a significant opportunity for them.