Microsoft is continuing its efforts to simplify its product lines, and has cut the number of versions for the next release of Windows Server down from twelve to just four: Datacenter, Standard, Essentials, and Foundation. Pretty straightforward. Windows Home Server gets axed, but then again, I don’t think anybody bought Windows Home Servers anyway.
It is odd that they impose those strange, arbitrary limitations on the number of virtualised clients, and registered users.
To me, it makes no sense to buy any of those, given the limits, when it is possible to get fully-functional, fully supported server software for far cheaper elsewhere.
Does the windows logo still mean so much, even these days?
Well in my company we are migrating a lot of projects from UNIX to Windows.
Why? I can’t see any advantage MS-Windows would have over UNIX.
If you can not find a single reason to move from original UNIX (not GNU/Linux) to windows you better hit the books my laddie =D
I have a lot of money and I don’t know where to spend it. I will switch from Unix to Windows to pay a lot of CAL licenses.
Maybe that is one reason.
Edited 2012-07-06 02:55 UTC
“I have a lot of money and I don’t know where to spend it. I will switch from Unix to Windows to pay a lot of CAL licenses.”
Windows licenses are cheaper than commercial UNIX ones.
In most multinationals the amount of money spent in licenses is irrelevant when compared with the total operation costs.
Edited 2012-07-06 05:36 UTC
“If you can not find a single reason to move from original UNIX (not GNU/Linux) to windows you better hit the books my laddie =D”
Quite right, when I visited some of our customers it felt no different than using a PDP-11.
Many commercial UNIX boxes are using the default tools from System V days.
Back in 2000, when we were doing Solaris configurations, the first thing after getting the system running, was to install all the GNU tools from Sunfreeware.
Edited 2012-07-06 05:37 UTC
We are doing for our customers.
My employer does consulting for multinational corporations.
If you compare UNIX installations (Solaris, Aix, HP-UX) to Windows, it might not be worth it to keep them around.
These are usually the type of reasons we hear:
– Windows licenses are cheaper than commercial UNIX ones
– Many UNIX boxes are proprietary hardware
– Java can run anywhere
– Some are actually porting applications from Java to .NET
– Sometimes those systems are just legacy and what they are doing can be easily done in Windows as well.
– Cluster management tools are much better in the Windows world
– Some systems are not worth to move to UNIX compatible systems, because they need to be rewritten anyway
– Linux/BSD are evil (due to open source)
– Some dude in upper management decided to do it
If you leave out Linux, then yes. However…
Now compare to Linux – no license fees. Just a minor support fee which equals the yearly support for Microsoft any way, and is no worse up front than the initial server license without any CAL licenses. Or, of course, you could forgo the support fee altogether, but those big multi-nationals like the support train.
And Linux will typically use a lot of that same hardware if you desire (especially if you get the support from the vendor – HP, IBM, etc.) or you can move to cheaper boxes just like with Windows. And guess what? Nearly all server class boxes that run Windows equally run Linux.
Not necessarily, especially if someone wrote a path incorrectly for loading a file, accessing a configuration file, etc. Minor bugs to the Run-Anywhere method, but sometimes Java gets tied to the OS too. Just saying…
So going from a language that performs equally on diverse platforms to one that may perform better on a single platform whose owner only wants you to be on that platform so you’ll have a larger cost to move off later…yeah, that makes sense.
And those legacy applications can probably be moved over to Linux for less.
Here’s a good one. Linux Cluster Management is world-class. There’s a reason Linux runs on nearly all super computers, and clustering is one of them. Windows Cluster Server is nothing in comparison.
Except you can minimize what you rewrite, or just move them as they are. IBM, HP, etc. all have the ability to help you move those legacy applications to Linux.
Well, They’ve probably been running BSD for a while, but you can blame the MS FUD engine on that one.
And here’s the difference. Instead of listening to their employees they are instead listening to the marketing drivel and making top-down decisions influenced solely by outside factors instead of bottom-up decisions that take into account the actual needs of the organization.
Of course, the person that made the decision won’t be there to see the project fail, and the company spend twice as much to do the next iteration which will likely fail too. The position probably turns over in 5 years or less, and these projects take 6 years or more.
He already said it is cheaper, they obviously looked at it for their needs, not what you reckon it would cost.
New hardware and more expensive Redhat Support, no.
The same argument can be made between any migration, I have come upon many things such as hard-coded paths in PHP applications that have stopped things from running.
You just contradicted you last point. Also there is Mono.
There is no proof of this and it is purely conjecture.
He is not doing super computing. It depends what you mean by cluster, Microsoft Server Farm tools are pretty good … not all of them are supercomputer 😉
He knows more about those apps than you do, maybe the code is just so legacy it isn’t worth porting or maintaining compared to the price of a rewrite. Linux isn’t magically the answer, and have you seen IBM consultant fees.
That doesn’t make anysense. There maybe legal issues offering solutions that have GPL source in them. You don’t know the circumstances.
More conjecture.
Conjecture again.
Edited 2012-07-07 19:46 UTC
True. However, he didn’t break down into the why, and what I listed is how Microsoft does licensing.
For instance, here’s the formula:
Cost = X + C(M)(N) + S(N)
Where X = the cost for the license of the server OS, C = the cost of each Client Access License, M is the number of clients, N is the number of years for the system to be used, and S is the cost of each year for support.
For Microsoft you can use the formula as is. For Enterprise Linux versions such as RHEL and SLES it becomes:
Cost = X + S(N-1) = S(N)
As X is basically the cost of 1 year of support which is free for that first year (typically).
This is, of course, what Microsoft doesn’t want you to know. They can’t compete on price, so they try to make it out to be more about the TCO (Total Cost of Ownership) where they muddy the waters to make it look like they are actually cheaper when they’re not.
If you are purchasing new hardware from a Tier 1 vendor – which is basically HP, Sun/Oracle, IBM, Dell, and a couple others – then there will be support from Red Hat and SuSE as these systems get prime support up-front and they all have developers working on Linux to integrate their hardware offerings.
If you’re looking to put together a large enterprise system and you are not using a Tier 1 vendor, then you need to re-evaluate who is doing your purchasing/etc as they are likely an idiot.
True.
Mono is for all intents and purposes a dead-end. No one is supporting it any longer, not even it’s founder. They’ve moved on.
And Mono was never really a viable alternative to start with – why? Because it only represented the core, and 90% of the functionality used in .Net applications would never be available.
Mono is to .Net like using C++ and saying you can never use a Class.
If there’s anything else you mean by this, then please explain – but Mono does not solve the issue, and .Net runs only on Windows, so they are indeed going from a multi-platform language (Java) to a single platform language suite (.Net).
Those legacy applications are most likely POSIX complaint, which Linux does support. The Tier 1 vendors do provide wrappers functionality so that an AIX/HP-UX/Solaris/etc application can run on Linux as the return values and behaviours can be slightly different where POSIX is ambiguous. That is well known, not conjecture.
The wrapper functionality makes the porting to Linux very easy, and cuts costs substantially – in many cases it just runs as the intent is to use them on binaries that can’t be ported.
Just because you are not doing a Super Computer does not mean you are not using the same tools, and the various tools for cluster management of any size are far superior to what Microsoft provides.
Yes, Microsoft has come a long long way in their support for clustering computers since 2000. But it still does not compare to what other system have.
True, but did he get the right information from the right people? Did they do the comparison correctly? From what is shown in the thread, it’s not likely.
True, Linux isn’t magically the answer. True too IBM’s fees are high. But we don’t know if he’s dealing with IBM, HP, or another vendor. For all we know he could be dealing with Microsoft’s Xenix as the original Unix system – in which case MS won’t offer any help moving to Linux, or if they do they’ll make it significantly cheaper to go to Windows. (Yes, Microsoft owns their own official Unix variant.)
So there’s not enough information to say more than I did.
You obviously misunderstood what I said.
BSD is a Unix Variant. For him to say that BSD was not an option is like saying that Unix is not an option. FreeBSD may be open source, but AIX/HP-UX/etc all had incorporated some of the BSDi Unix from which FreeBSD/OpenBSD/etc were derived. So in all likelihood they were running BSD variant of some sorts already.
No. He explicitly stated it was a top-down decision made by the upper echelons of management. That is exactly who Microsoft markets to, and who they purposely mislead.
So the CEO/COO/etc who made the decision likely made it based on Microsoft’s marketing materials instead of consulting their employees to get a real answer which often would be very different.
For instance, many large organizations are already running Linux even if the upper management doesn’t know it. It is very possible that they were already able to run their legacy applications on Linux without any costs for porting, but they didn’t consult their own employees and thus didn’t find out that that may have been the case – in which case, they already have all the in-house knowledge to move off the Unix systems to a Linux-based system, no need for retraining costs, etc.
So did they really make an informed decision? We won’t know.
[/q]
No, that’s how companies have been running for the last 30 years.
TBH, I was putting it in there because I was being trite after the “Java Doesn’t Run Anywhere” comment, when before you said “Should be portable between nix systems”.
“Most Likely” … so it is still a guess.
Considering some of the cluster-fuck coding I have seen in C#.NET by contractors (who seem to think it is Java) and considering C# and Visual Studio should make things bloody easy to write nice code … well you see where this is going.
I know how stuff works thanks, I might be a lowly web dev, but I did OS 101 stuff as well.
If the application is really old, chances are that you could probably do a rewrite using C# for much cheaper as long as you get the requirements right.
Why do you guys seem to be insistent on schooling me about where various *nix came from? I bloody well know where it came from.
Exactly you don’t know, so conjecture.
It was just one of many scenarios. Just dismissing many other valid reasons given, as “not likely correct”, is even worse than conjecture…
Let’s remember that they can start using GPL software without even having to read the GPL.
You are not required to agree to anything to merely use software which is licensed under the GPL. You only have obligations if you modify or distribute the software.
— https://www.gnu.org/licenses/gpl-faq.en.html
That can still stop or complicate “hey, that software we were doing for our needs turned out good, maybe we can sell it to others”.
Or, in the volatility of corporate setting, can easily bring headaches with spun off divisions and such.
Don’t look at it like Free Software advocate – corps adopt it more or less only when they can exploit it to their benefit.
“Datacenter, Standard, Essentials, and Foundation.”
If I don’t check the MS page, I can not even guess what is the difference between Essentials and Foundation.
Just like the two IBM Lotus Notes types “Basic and Standard”. …should they call it “basic” and “plus” to understand which offers more functionality?
Edited 2012-07-05 21:57 UTC
Windows Home Server has never been on the same release cycle as the business servers. It usually comes months later because they build additional functionality atop the base server. Unless there is word from MS, it’s premature to say it’s gotten the axe, especially when Storage Spaces provides a superior replacement to Drive Extender, and there are a number of improvements to manageability, and opportunities to tie into SkyDrive and other cloud services. They may, indeed, decide not to make a new version, but they’ve got the technologies available to make the best one yet — even offering more embedded, appliance-like options.
Update: Home Server has actually been rolled into the Server Essentials SKU.
http://www.neowin.net/news/windows-home-server-is-no-more
Edited 2012-07-06 00:02 UTC
But I saw that one coming. Windows Server for a low price of 50$? Too bad WHS2011 was just an average release. No efforts were made to tailor it further for consumers (other than proper DLNA support).
I’ll be moving my WHS2007 to Windows 8 Pro. Storage Spaces is fantastic and just what I need (Drive Extender on steroids). Remote Desktop is better then having to install WHS Console on select PCs.
Just need cheap mobos featuring “Centerton” Atoms, ECC RAM and plenty of SATA3 ports.
If I have an ASP.NET application that needs to run on a server in a organization that does not already have any licenses, what do I need to buy?
Lets say the org has 100 users and the app does not need AD or anything beyond IIS itself.
The app is taking care of authentication itself (no user accounts needed) just straight up ASP.NET.
From what I can see depending on the versions:
alt 1: 1 DataCenter Edition License + 100 CALs
alt 2: 1 Standard Edition License + 100 CALs
alt 3: 1 Essentials Edition License (no CALs needed)
alt 4: 1 Foundation Edition License (no CALs needed)
Is this correct? They removed CALs from Essentials and Foundation?
For a web application that does its own authentication and does not call Active Directory, you do not need to buy any CALs at all.
IIS is considered to be a “specialty application,” for which CALs are not necessary.
Windows Web Server never required CALs, either. CALs are really for Windows services — file sharing, printer sharing, remote desktop, Active Directory, etc.
Edited 2012-07-06 05:02 UTC
1
Browser: Mozilla/4.0 (compatible; Synapse)
1
Browser: Mozilla/4.0 (compatible; Synapse)
1
Browser: Mozilla/4.0 (compatible; Synapse)
1
Browser: Mozilla/4.0 (compatible; Synapse)
1
Browser: Mozilla/4.0 (compatible; Synapse)
1
Browser: Mozilla/4.0 (compatible; Synapse)
1
Browser: Mozilla/4.0 (compatible; Synapse)
1
Browser: Mozilla/4.0 (compatible; Synapse)