The age of six-monthly Windows Server updates starts … now!

Microsoft's vision of six-monthly Windows Server updates is now a reality. Announced just four months ago in June 2017, Redmond's new Windows Server plan will see it offer a “Long-term Servicing Channel” that behaves a lot like Windows Server does today: a big upgrade every three years or so and support for more than a decade …

[The Original Steve]

Upgrading every 6 months?! No chance. But...

I must say the new hyperconverged web UI looks very tempting...

Had some excellent results in using Win 2016 as Hypercovereged platform so far. HyperV 16 and Storage Spaces Direct are actually really quite impressive for the same price we'd pay for Win Datacenter anyway for HyperV hosts. Cost for deploying hosts and storage have gone through the floor as a result.

Where's it's been lacking has been on the management tools side of things, presumably to get people to shell out for Sys Centre which unless you have 2 years and a deep loathing for yourself, your time and your life isn't viable for all but the largest of estates. Good to see improvements in this area.

Although long term the effort to upgrade every 6 months does dilute the benefits somewhat...

[TheVogon]

Re: Upgrading every 6 months?! No chance. But...

"The Register lacks credentials to Microsoft's Volume Licensing Service Center where it can be found."

It's on MSDN too. Confusingly it's labelled just as Windows Datacentre Server and you have to dig into the description to find reference to build 1709...

[Steve Davies 3]

Sigh...

One of the systems I worked on needed around 6 months of integration and testing before an OS upgrade. All the different 3rd parties had to buy into the work. Some were really reluctant to even do any cursory testing. We did OS updates every 18 months. These were systems that were air gapped from the internet btw.

I forsee lots of Admins just sat 'Meh' and carry on with the annual updates if that.

Once a system is running properly the old saying, 'If it ain't broke then don't fix it' rules.

[Anonymous Coward]

Re: Sigh...

Annual?

Is that the same as once every 10 years or when it breaks...whichever comes last?

[Anonymous Coward]

Re: Sigh...

"I forsee lots of Admins just sat 'Meh' and carry on with the annual updates if that."

That's the whole idea. You have the choice. You only deploy these non LTS versions if you want to use newer functionality. And even then you can choose to wait until the next LTS release to get it...

[David Lawton]

We already have a few Linux Server boxes (Web Servers, Asterisk VoIP, DNS Servers) but mainly Windows. Think its time we start seriously moving more to Linux now unless we HAVE to run Windows because of Server Software requirements.

The update merry go round is making me sick.

[Anonymous Coward]

You do realise that Linux get update often as well?

[Pascal Monett]

Yeah, but it gets broken a hell of a lot less.

[Anonymous Coward]

"Yeah, but it gets broken a hell of a lot less."

Speak for yourself. We get way more problems from patching and updating Linux boxes than we ever did from Windows Server ones.

[Anonymous Coward]

You do realise that Linux get update[sic] often as well?

Depends what you mean by 'update'.

Linux gets frequent updates to fix bugs. It also gets infrequent upgrades, which add new features and which change the way the system functions.

Updates need to be frequent, or at least as frequent as the emergence of the bugs they fix, whereas upgrades need to be infrequent to allow time for testing before production deployment.

A six-month update cycle is too long to wait for fixes whilst a six-month upgrade cycle is too short for adequate testing of production systems.

[jonfr]

If you are looking for easy time, it is wrong for you to move to Linux. If you want easy time you need to move to FreeBSD or NetBSD. You also have the option of OpenBSD. It is far easier to maintain and update FreeBSD than any Linux version that I've used so far. I have not yet used NetBSD but it is on my plan to do so (for web service or something else useful).

[Anonymous Coward]

Can't see its value beyond developing/testing

Unless you need some very specific feature available only in one of these "fast releases", I can't see really why you should risk unwanted effects. For development and testing OK, you know you'll run on the last release too, but most production setups needs to be stable, not fancy.

[Fred M]

As it's optional, surely it's a good thing? Most people will quite rightly stay on LTSB. However, if there's a feature that you want (and is worth the regression testing) then having it available promptly on the 6 monthly schedule is a bonus.

[Franco]

No issue with it for me as long as they keep the long-term support on specific builds as they have with client versions of Windows. Not sure of the point of it personally, given that most companies will want their business running on long-term support.

[Anonymous Coward]

I'm still indifferent...

I don't think this has to be a bad thing to be honest, as long as it is optional. To be honest I'm a bit confused because it almost sounds as if MS will drop support after one year or so, forcing you to upgrade no matter what.

Even so, I'm still indifferent. We replaced our Windows server with FreeBSD, Samba, Apache, mod_mono and PostgreSQL several months ago and so far so good. Heck, instead of having to get extra (heavier) licenses in order to utilize a VPN all we had to do was "make -C /usr/ports/security/openvpn install clean", configure the whole thing and wham.

Of course the downside to all this is that hardly anything exciting happens anymore during updates :P

[Anonymous Coward]

Re: I'm still indifferent...

Isn't it same as Linux builds?

"Normal" and "LTS"?

[jbuk1]

Re: I'm still indifferent...

I don't know of any Linux distributions which have only 6 months between releases regardless of LTS or not.

[Anonymous Coward]

Re: I'm still indifferent...

"I don't know of any Linux distributions which have only 6 months between releases"

Fedora?

[Pascal Monett]

"while Azure users don't notice OS changes"

I'm sure they will.

Soon.

[ab-gam]

Server is not xBox

Our IT needs are simple and fairly static. Barring any externally forced changes, I see no value in straying from our standard process - Buy new hosts with a current OEM copy of Server, add security updates for 5-7 years, lather, rinse, repeat.

The drama of twice yearly update risks is not worth whatever tweaks MS decides we need.

[JM987]

I like the idea

I believe most of the comments so far are from people that deal with infrastructure and not developers that want to have access to the latest tools. From an infrastructure perspective, upgrading and migrating operating systems every 18 months is way to short. From a development perspective, I can see the huge benefit as I often work with developers that want to work with the latest products available. If this means we deploy a 18 month release for non-prod environments and they get an opportunity to develop applications and install them in a LTS release that will be coming down the pipe in a foreseeable future, that could allow us to get ideas formed and projects started much earlier.

It has its place and will eventually become the new norm once it catches on.