Ask HN: If WinServer2019 was made free to use tomorrow,would you start using it

35 points by ThinkBeat ↗ HN
If Microsoft decided to make using Windows Server 2019 (all edition) free to use tomorrow, would you start using it? And if so how?

Would you be more likely to adopt it if Microsoft open-sourced it? (I know very little chance of that)

In either case Microsoft could make money on support, and consulting services.

I think more competition on the server side, than various flavours of Linux and some flavours of UNIX would be healthy.

For a time I worked in Windows only enterprises and I had little to do with administering any of the servers, but code got deployed to them and I never had any issues with the parts of the server I used.

I would think that there are some workloads and tasks that might make more sense to run on a Windows Server vs a Linux one but I am not sure what (aside from legacy Windows only applications)

65 comments

[ 3.6 ms ] story [ 105 ms ] thread
This is something I call a “wish in one hand, sh*t in the other , see which fills up faster” scenario. There is no way Microsoft could make as much as they do now on Server if they switched to a “services and consulting” model.
Also currently there is a bit of an incentive to make it user friendly.

If it becomes "consulting-ware" like Linux there would be an incentive to make it user-unfriendly and complicate it with high costumizability, bad UI and bad defaults.

What would it buy? Unless the org suddenly switches to a different tech stack that requires winserver, it’s kinda the worst of both worlds - none of the upside of Windows, none of the huge open source ecosystem of Linux.
So far, the only use I've had for running Windows machines was running Adobe After Effects for a render farm, because you can't run the Adobe suite on Linux. I struggle to imagine workloads I'd want to run on Windows machines voluntarily honestly, but I'm also unfamiliar with the ecosystem from a sysadmin perspective.
I might use it as a desktop, or for servers that had to run Windows-exclusive software, but I wouldn't go out of my way to use it if Linux was available as an option.
Active Directory is a blessing compared to the hellish OSS alternatives. Don't reply with suggestions to use FreeIPA. It can be just as bad or worse.
Samba is pretty decent
Thats like saying tires are a decent replacement for Uber.
How? Samba supports AD, and can work as a replacement for Windows AD (or together with Windows DCs)
You don't need Windows to use Active Directory.
Oh? Please elaborate. I always assumed Active Directory = Windows.
Doesn't AWS have it as a service? Possibly also an OSS project?
Decently common to have a cloud AD instance for your internal “users table” even if all the use cases are web/Mac/Linux.
Active Directory is a Microsoft product, which, last I knew, only runs on Windows.

Active Directory implements LDAP, and there are serveral other LDAP implementations. Other OSes can use Active Directory as an LDAP server, sure.

AD is really powerful, but as someone who accidentially slided into Windows administration in his job, I find the windows Admin tools lacking the most elementary features I assumed they would have:

* Quickly get a shell on a remote machine, like I'm sitting in front of it? Powershell is nice but it's remoting model is different. You can't easily manage Windows updates for example over Powershell, it just behaves differently. * I assumed there would be something like Win98 "network neighbourhood" on steroids. Give me a list of all computers in my domain, or all computers in a group, or all computers certain users are on. AD only tells you which ones are joined, you cannot browse which ones are actually running. * Some GUI where you can ping a computer, find out which user is logged in, which software is installed, and which updates are missing.

You can find nice third party tools for all of this, or spend a lot of time writing scripts, but the basics are lacking. And the offerings from MS (WSUS, SCCM, ...) are powerful, but so far would cost us more time than they save, so I'm not really happy with the Windows ecosystem...

I might be missing something, but what makes Active Directory so much better than any other LDAP implementation?

My limited understanding is that it has better tooling and defaults, and that it's more than just an LDAP implementation, but is it really that much better if you aren't in the Microsoft ecosystem?

I've recently come across a situation where somebody couldn't log in with their AD credentials, turned out the account had been renamed... and nobody knew who (or what) did and why.

I don't know if AD comes without audit logging, or it was just disabled in that installation, but that seems to be a pretty big no-no for a central part of your authentication infrastructure.

(Granted, it's an anecdote, but too many WTF anecdotes make me not like a thing as well).

I think AD is probably the best (if not the single) reason to use MS boxes. It's quite a beast and you have to know its little tricks but hey, the same goes for any other solution of similar scale.
(comment deleted)
Personally? No.

I gain nothing from Windows except overhead, particularly when it comes to servers.

I generally pare my servers down to little more than the kernel, a firewall and my very small handful of binaries for my application.

Microsoft made their consumer OS basically “free” and in doing so loaded it with so much telemetry, data collection, bloatware, and dark patterns, that it is unusable without significant and ongoing intervention.

So, no, I would not want them to have to find alternative income streams from their server OS.

Are all of these things included in the versions of Windows they make available on Azure..? I wondered that, but wasn't going to make an account just to find out.

My kid wants a laptop for gaming... he's been using my MacBook Air for KSP and Minecraft for a long time, and I won't be buying a new machine until the new models are out and proven... so I've conceded to getting him a gaming laptop for the time being.

It'll be Windows. It'll probably be our only Windows machine for the foreseeable future, but I know we'll have to de-clutter it when it gets here. And I still need to research that.

Is there a HN approved way of doing this?

I recall there being various solutions that cleaned older versions of Windows, but perhaps did so a bit too aggressively (disabling services left and right in the XP era), so experts recommended against using them. I have no idea about the current state of affairs, but may find myself in the same boat soon and need to know.

> Is there a HN approved way of doing this?

Yes. Not using Windows to begin with.

Getting windows signature devices is a good start. It doesn't get rid of the baked in telemetry, but it does ensure you're getting a fresh copy of windows (with Candy Crush and whatever store apps they decide to install) without any of the bloatware OEMs like to install.
My windows laptop runs the LTSC edition (likely in violation of the licensing terms) which is meant for super strict environments like ATM and POS machines. It does not get user experience updates and comes extremely barebones - but to me it Works exceptionally.
>loaded it with so much telemetry, data collection, bloatware, and dark patterns, that it is unusable without significant and ongoing intervention.

It's very much usable for everyone that ignores telemetry and bloat, which is most people. Telemetry gathered likely makes the experience better for people.

Approximately 30GB footprint for WIN10 vs approximately 1GB for Linux.

That some can 'ignore' this fact (which has other similarly astonishing dimensions) does not make it irrelevant -- the impact/implications are real.

Windows typically gets installed with all the supported drivers. NVidia's blob alone is over 600MB.
a 30GB install does not make it "unusable" on most machines.
Same for the Linux kernel. It includes all drivers.
It was temporarily free. Buying a license now is $100. I use Macs, but I've been wanting to buy a Windows license to run in a VM, and I couldn't believe how expensive it is.

With the amount of data collection and ads in it, it should be free.

I administer a few Windows Servers. Unless you're heavily in the MS ecosystem (Azure, AD etc.), or have software that only run on Windows, I can't think of too many advantages that Winserver has from a server management perspective. Maybe the admin GUI?

I understand that even on Microsoft Azure, most of the VMs being hosted are running Linux.

https://build5nines.com/linux-is-most-used-os-in-microsoft-a...

It depends on the use-case, but probably no. For most use-cases I can think of (Active Directory), the cost is already negligible for a business.

What use-cases would be attractive for free versions? I can think of one, VMs. But that experience already sucks. Windows images are larger, slower to boot, and harder to automate.

It could be quite useful for CI though. And more CI providers might start offering Windows platforms if it became free.

I know great strides have been made in recent years, but Microsoft would have to communicate a much better 'configuration automation' story to significantly move the voluntary adoption needle for this crowd.
The TCB would need to be open-source, e.g. Hyper-V, MinWin (or whatever the smallest core of windows is called these days). There are several virtualization-based security features in modern Windows, that would be especially useful if the hypervisor could be inspected and collaboratively developed, even if a Microsoft CLA were required for code contributions, and they retained patents. An OSS TCB would ensure absence of telemetry. They can still sell closed-source components that run on the OSS TCB.
No. If it had something you needed, and that you could only get from Windows, you'd be using it already. Any shop that can make use of an alternative pretty much already does, and would have no reason to switch.
As a decades long FreeBSD and Linux user, I can say no, I would not use Windows on a server come hell or high water for jobs that are already better served on open source platforms. It is seldom the best tool for the job and Microsoft's own metrics show that Linux will become predominate even in Azure.
No.

.net core runs on Linux. Why would I invest in a new Windows server?

The cost is stepping in the Microsoft tarpit. No chance in hell I'm getting near it.
No. Even if open sourced, no. There would be no benefit to me in doing so (client or server) as it only has value for Windows users, which I no longer am. I was a Windows user long enough to be able to say with a fair degree of certainty that its code base would have little that I'd be interested in.

Look at it this way: if IBM released for free, or open sourced, a port of its mainframe OS for PCs how many people would start using it? Unless one already had a significant investment in IBM mainframe software, probably not very many.

I'm all for competing ideas and alternate Operating Systems but don't view commercial platforms as the place to look these days for good ideas/implementations.

This is a good point. Most people when comparing the popularity Linux vs Windows Server or Apache vs. IIS ignore the fact that Windows Servers costs a bunch. A proper comparison would be if both were free or cost the same. I wonder how the server wars would have turned out if that were the case.
No. The point of Windows Server at its peak (around 2000) was that it was an easy to manage server OS.

I don't want to manage any servers. No boxes, no kubernetes, no anything. I have have some code, I pay my cloud provider for an execution environment and that's it.

It might cause me to upgrade current windows servers instead of migrating them to Linux. But I doubt I'd start new projects with it.

If mssqlserver also became open source then that might change my mind.

I've been using Linux for a long time, Windows servers have only been a small part of my experience. At one point I decided I was going to invest more in the open source side of things, I learned Linux / Python / Bash / Go / Nginx and Apache, etc. I can use .NET and Windows server but I prefer not to. I guess at this point what I'm saying is, it's too different from what I've invested my skills in. I'm going to guess that others probably feel the same way. Linux has huge mindshare in this space and making Windows server free is not going to convince people to switch.
We have a datacenter license which isn't really that expensive, relatively speaking. Windows isn't the expensive part. It's everything else.

Our SAN is expensive. Easily 10x the cost of the raw storage.

VMware is expensive. Backup software, management software, Exchange, our wireless network solution, our MDM solution, that's all expensive.

RHEL isn't free either.

You are not kidding on the price of SAN.

I just started a pilot program with the iXsystems TrueNAS, which is the supported enterprise version of FreeNAS.

It was easily a 10th the price of everyone else I looked at. Plus the support contact was inexpensive.

Now if I could just figure out something about our VMware licensing. I've looked at RHEV, but we have a lot of automation built around VMware. It will be very expensive to switch.

The SAN would be equally expensive in any OS context would it not?

The few times I have been peripherally involved with acquiring on-site enterprise SAN storage the prices have been out of this world but not every dependent on what the OS accessing the thing is

Very small (less than 20 people) non-tech businesses, especially those with on-site culture can actually benefit a lot from such a deal.

The management UI's are actually useful and almost a must have for many people.

Simple answer, No. Microsofts overall approach to licensing is complex enough that you have to be an expert in their license to even have an inkling of an idea what is permissible and what isn't. The OS layer is just the tip of the iceberg. It's not immediately useful in and of itself without things like SQL server, IIS/Asp or whatever their web app layer is these days, etc. The would have to shift gears on pretty much every technology to make it worthwhile for a Linux/*nix stack developer to look in their direction.
MS cannot even explain their licensing at any detailed level unless you are speaking with one of their own 'Licensing Experts'.

Not exaggerating -- this has scenario has played out again and again in my dealings with MS going back almost 30 years.

I probably would have a few years ago, now-a-days almost everything I do seems to be deployed as a linux container directly or linux containers by way of 'cloud functions'

funnily enough I use windows as my workstation os