Windows 10 May 2020 Performance For WSL vs. WSL2

[torsionbar28]

I'm afraid that Microsoft will soon introduce something to try locking developer inside WSL so they'll start deploying WSL on the servers (or just make it very convenient).
As much as I hate Microsoft, I have to recognize that their strategy at the moment is actually damn good.

Pardon my ignorance; I don't have any hands on experience with recent Windows server versions or with WSL. That said, what is the use case for WSL on a server? I thought WSL was more of a Cygwin competitor, something that allows a Windows PC to run Linux binaries. Like WINE, but in reverse. If that's true, then I cannot imagine anyone hosting servers with it.

[anarki2]

So these results mean that developers will now stay on windows and use linux within it and just deploy to linux servers which will further destroy desktop linux marketshare?

To begin with, desktop Linux market share has never been outside the realm of margin of error. There's not much to destroy.

[torsionbar28]

To begin with, desktop Linux market share has never been outside the realm of margin of error. There's not much to destroy.

It's mainly software developers, IT types, and scientific workstations that run Linux. Plus the enthusiast/hobbyist crowd. Historically not much different from proprietary UNIX workstations. It actually is a significant market, but as you pointed out, is a tiny fraction of the whole when you factor in business desktops and mass market home users.

[cynic]

Pardon my ignorance; I don't have any hands on experience with recent Windows server versions or with WSL. That said, what is the use case for WSL on a server? I thought WSL was more of a Cygwin competitor, something that allows a Windows PC to run Linux binaries. Like WINE, but in reverse. If that's true, then I cannot imagine anyone hosting servers with it.

I don't have experience with that too (luckily ) but I guess that the server is the aim of microsoft for WSL, there's no other reason why they're integrating Linux inside Windows (they already dominate Desktop).

What I'd do if I were Microsoft, would be create an ecosystem that allows developer to easily deploy their "linux" application from their windows workstation directly to their Microsoft server with just one click and that allows them to monitor and manage that software as well.

That would create an incentive to use the microsoft "hybrid beast"on the server instead of another enterprise Linux distribution and honestly, I think it will gain a lot of marketshare.

Plus, they could "extend" their Linux with some special feature, as some kind of integration with the Microsoft world.

That would be their classic Embrace, Extend and Extinguish strategy.

[CommunityMember]

So these results mean that developers will now stay on windows and use linux within it and just deploy to linux servers which will further destroy desktop linux marketshare?

Half of nothing is, well, let me do the math, hmm, it is still nothing.

There may be many reasons the linux desktop marketshare is in the noise, but WSL(2) is not going to change that. And the year of the linux desktop will probably happen the same year that IPv4 becomes meaningless and fusion power plants become commercially viable (all of which are always decades away no matter when you ask the question).

[Tuxee]

Jeez. The conspiracy theory idiots are running rampant - and all because of a single benchmark.

[cynical]

So these results mean that developers will now stay on windows and use linux within it and just deploy to linux servers which will further destroy desktop linux marketshare?

That implies people were only using Linux for performance reasons, or that they would stand to gain from using Windows.

As a developer, I avoid Windows because I value privacy and control over my own hardware, something that Microsoft will never give me.

The fact that my system environment matches the environment of the server I might want to deploy to is a bonus.

[mb_q]

Overall, what was to be expected: with modern hardware, the impact of the OS, and even VM, on a CPU-bound program is negligible. It would be nice to also have vanilla Windows and Windows in KVM on Ubuntu for comparison...

The lack of any difference between WSL2 and bare-metal I/O somewhat suggest that Windows is not properly relying sync operations to the VM image, though.

[duby229]

It's mainly software developers, IT types, and scientific workstations that run Linux. Plus the enthusiast/hobbyist crowd. Historically not much different from proprietary UNIX workstations. It actually is a significant market, but as you pointed out, is a tiny fraction of the whole when you factor in business desktops and mass market home users.

Desktop Linux has at least 2% marketshare as a bare minimum. That's still millions of people.

[duby229]

Half of nothing is, well, let me do the math, hmm, it is still nothing.

There may be many reasons the linux desktop marketshare is in the noise, but WSL(2) is not going to change that. And the year of the linux desktop will probably happen the same year that IPv4 becomes meaningless and fusion power plants become commercially viable (all of which are always decades away no matter when you ask the question).

Again Desktop Linux as a whole has at least 2% marketshare at a minimum and that is still millions of users. It's far more than nothing.