Don't know anyone's ages here, but is anyone fascinated by the historical implications of this? I started using Linux in the 90's and the talk in the community at the time was "world domination". It seems to me that's happened--more or less behind the scenes, but it's happened.
I couldn't disargee more about Microsoft's current place in tech markets. They've lost quite a lot of weight since their 800lb gorilla days in the 90s.
They used to exploit their dominant market position to replace de-facto standards with their own: kerberos "extensions", creating ActiveX instead of just implementing NPAPI, removing JNI from the MS JVM. But they can't do that any more. They lost too much ground under Ballmer.
Now they're nearly tripping over themselves promoting their cross-platform tolerance. Run Linux in our cloud! Run our apps on all the platforms! Sync/connect our apps/devices to any service!
Source: Age 34, started with Linux in '97 (Slackware 3.2)
I couldn't disargee more about Microsoft's current place in tech markets. They've lost quite a lot of weight since their 800lb gorilla days in the 90s.
You are aware that the "World Domination" being referred to was talk of Linux's World Domination, right? So you're not disagreeing, you're supporting the observation that Linux eventually dethroned the 800 lb gorilla and that the 800 lb gorilla is a shadow of what it once was.
For all the Old People, replace Microsoft with IBM above and transpose this back into the 1990s.
IBM went from EBCDIC and Micro Channel Architecture and SNA to "Well, they'll never actually beat Compaq and Dell in the primary growth market, but they still own the tippy-top of the Enterprise world. That's something."
Oh, and about this time, IBM shifted focus from AIX to getting behind Linux in a big way. Remember the TV commercials?
37, started using Linux in '98 (what can I say, late bloomer), but ran OS/2 before that as "rebelling against the M$ hegemony" (plus OS/2 was technically superior). I still have some of the IBM Linux ads.
I still don't trust Microsoft and even if I did, I consider most of their offerings piss-poor and only "industry standard" because the "industry" is run by upper management idiots. (note that the "industry" I refer here to is most definitely not SV or startups; think old, entrenched organizations that have no impetus to change or improve. You know, like government orgs. Or Microsoft).
40, started using Linux in '93 or '94. You're probably old enough to realize that even though you're kind of in the same casino, those "upper management idiots" are playing a vastly different game than you or your startup. What portion of your work, for example, is spent figuring out whether the EU is going to give you a seven-figure fine for not localizing something, or whether a vendor from a non-NATO country can troubleshoot a server with a given client's PII on it, or whether the update you're pushing this week has been tested to make sure none of the 27 major products that take dependency on it are going to break? If your only metric is how much chrome the end user sees...
Yeah, it's so surprising that Microsoft didn't implement the "Netscape Plugin Application Programming Interface" (which was just for web browses) or other interfaces that weren't build for Windows.
Not only was NPAPI developed by a rival company, it was specific to browsers. ActiveX allowed features from any (Windows) application to be embedded into any other (Windows) application, not just browsers. It was a strictly more general framework and thus a better option for Microsoft. What really screwed things up was, security was not one of the foremost considerations when designing ActiveX because it was conceived for the desktop world. So it was horribly insecure when applied to arbitrary content downloaded over the web.
I'm 35, started using Linux in '97 so I've been around for most of the history.
It's quite amazing how Microsoft has changed in that time tbh from "Embrace, Extinguish, Extend" and all the shenanigans they pulled with comparing Win2k vs Linux/Samba and the halloween memo through to submitting code to the kernel, using Linux as a guest in their VM stuff and now using it internally not to mention making .Net a first class citizen (which when it's done will be incredible, I love C# as a language but in Linux land it's not quite their yet compared to on Windows).
MS used to really trash open source, with Ballmer calling GPL "cancer". (Although the company pulled back from that extreme).
When MS took over Hotmail they got rid of FreeBSD that was running it.
But MS have used Unix for years. They released MS IE for Unix. (The only person I know who ever used it was on HN.) Perhaps more bizarre was the Windows Media Player for Unix. Their TCP/IP stack was from BSD for ages (and possibly other stuff. The EULA used to refer to the regents of Berkley.)
I downloaded and played with IE for Solaris, but I didn't use it for day to day browsing. One interesting thing that I recall from it was that it had .so files that corresponded to each of the windows DLLs. So it looked like they had some sort of framework for porting windows code to Unix.
There used to be a page from the IE for UNIX porting team[0] that talked about the challenges of a direct port. Last I checked, only a copy of the content could be found on a mailing list post[1].
There's some interesting stuff in there. This one always seemed like a weird problem:
Global variables in Win32 dynamic-link libraries (DLLs) are visible across
processes, so it is important that developers are careful that these globals
do not conflict with any variables within the Unix shared libraries.
At least in the way it is presented. What does variables in DLLs being visible across processes on Windows have to do with variables in other libraries? Of course they need to be uniquely referenced.
I have misgivings about GPL. It's given me a lot, and I appreciate that, but while developing I won't use GPL libraries because I simply don't like being told what to do, and per the golden rule, I don't like telling people what to do, so I cannot with good moral standing release GPL code.
Well in the case of GPL, what you are not allowed to do is to use GPL licensed code with yours and decide not to give recipients of the resulting code the same rights that you were given.
I personally think that makes for a decent 'moral standing'.
Basically GPL only a problem if you want to create proprietary software, which in turn is typically very much a 'being told what you can do with said software' type of situation, with no code and often incorporating DRM mechanisms to prevent copying.
The vast majority of my code is either in scripted uncompiled form, or statically linked. Wordpress implied a while back that using uncompiled GPL libraries made your code GPL. I'm no license expert, I'd rather just not risk it.
The more commonly used term is viral, which these days, I think captures the can-be-the-right-choice/can-be-the-wrong-choice nature, while still saying it spreads to most things it touches.
The GPL is a cancerous license, and Ballmer had to trash open source. That was their business model and they were using the OSS association against Novell.
Edit: For those of you downmodding me for saying that the GPL is a cancerous license, please understand that this isn't a negative connotation - it's simply an aspect of the requirements of the license. It's very likely a good thing, and part of the reason that Linux is doing well while BSDs have languished.
It is not cancerous, it is just infectious. And not by accident, it is by design.
I used to like GPL, so nobody would "steal" my code adding it to a closed source product. I don't care anymore, just use BSD or other credit-where-credit-is-due license: my code was included in a high profile closed source product? Lots of bragging points and free publicity!
Hey, you're not wrong. Linus used the GPLv2 only because it allowed him to get patches back into the kernel. He's neither a fan of the FSF nor the GPLv3 because of their philosophy on software freedom.
Linux is by far more popular for developers, tinkerers, and cool software projects but BSD is still massive in the server and embedded devices space.
BSD is a minority in servers, which also reflect my own experience talking with other sysadmins in conferences. Debian and Redhat are huge in the server market, outside of windows shops. Some older system might have novel or a BSD system running together with massive amount of legacy perl scripts, but those companies also complain that they have a hard time finding new employees that can support the old systems.
The GPL is not a cancerous license, if I would use a negative adjective to describe it, it would be: misused.
I understand when people wants to keep making FOSS better by releasing entire projects with a GPL license, but I don't understand when people release simple stuff such as a code example with a GPL license...
The GPL is a cancerous license. If you use GPLed code, your code must (almost always) become GPLed, thus the GPLed codebase grows. If a GPLed project use other OSS code, it tends to be able to keep those changes, but the opposite can't generally be reciprocated (You can't take GPLed aspects of Linux and port them to a project under the BSD license without changing the license of that software).
This is not a bad thing. It sometimes sucks, but overall it's helped preserve the ecosystem and it has been integral, IMHO, in making sure companies stay honest with their upstream OSS meal tickets.
Way to quote selectively there. "It sometimes sucks" occurs more frequently for many people than many would like to admit. I work for a company that uses a ton of open source and open sources a ton of its own technology. Even for us, the GPL is poison. We're not allowed to touch GPL code. And we're not alone in this sentiment. A large number of open source enthusiasts decry the viral nature of the GPL, precisely because it has unpleasant properties common with cancer.
If GPL is cancer for a company, then what then is proprietary licenses? nuclear bombs? Toxic waste? Ticking time bombs?
I don't think such language is nice. It is damaging to civil conversation and can only alienate people more. No one likes hearing that what they create is like toxic waste.
How are proprietary licenses any better or worse than the GPL? It totally depends on what the terms of the license demand.
Note that it is not the code that these people produce that is toxic, it is what they demand in return (via the terms of the license) that is. The language may not be nice, but it is descriptive of the situation the GPL puts many people into. The GPL essentially seeks to impose the ideology adopted by a subset of folks onto the rest of the world. "We believe code should be free, so if you want to use our stuff, your code must be free too, regardless of what you believe or what your business needs are or what is practical." That is not a very nice stance to have, and so many people do not say nice things about it.
Think of it like someone saying, "Here, I'll do you a favor, but you must change your religion". Having different ideologies is perfectly fine, but enforcing them on others is not.
Don't forget Halloween Memo X, along with as I remember stuff from GrokLaw, which revealed that Microsoft funneled 86 million to SCO when it was litigating SCO v. The Open Source World.
(Especially bogus because SCO knew in advance they didn't have the rights to UNIX(TM), but I doubt Microsoft knew that.)
Huh, I don't remember that, especially if those were the two different SCOs. The first, the Santa Cruz Operation, was OK ignoring some quality issues, and they are the organization that Microsoft got 25% of through that transaction in 1987. Much later, in 2001, this happened: https://en.wikipedia.org/wiki/Santa_Cruz_Operation#Asset_sal...
And Caldera, which renamed itself in the next year to The SCO Group, got all of the original SCO's assets but Tarantella. Not sure what happened with Microsoft's ownership of the original SCO, but I don't remember through the years of SCO v. The World that Microsoft had a large ownership stake ... which would have been major.
There you have it. At least a third of SCO's entire market capitalization, and their entire current cash reserves, is payoffs funnelled from Microsoft. Their 10Qs reveal that every other line of cash inflow is statistical noise by comparison. The brave new SCOsource business model is now clear: sue your customers, shill for Microsoft, kite your stock, and pray you stay out of jail.
I also seem to remember some Microsoft related deal involving the purchase of Novell SUSE? licenses, although now I can't remember if The SCO Group directly benefited from it.
Microsoft earned this part of its bad reputation, and I'm sure I'm not the only one who's purchases from the company in part ended with Windows XP prior to that (of course, it helped that Vista was long delayed and awful). Supporting Microsoft back then was unconscionable.
In 1987 Microsoft sold Xenix to Santa Cruise Operations Group for a 25% ownership shared of Santa Cruise Operations Group.
In 1993 Caldera purchases Santa Cruise Operations Group, and renamed their collective company to SCO Group. Caldrea was part of Canopy Group which has a lot of subsidiaries.
Round about 2003 the SCO vs. Linux fight started. Microsoft still owned Caldrea stock, and provided them financial assistance as they attempted to sue the entire world.
Sorry in 2001 Caldrea didn't purchase Santa Cruise Operations Group outright. It just purchased everything relating to Unix.
After this sale Santa Cruise Operations Group changed its named to TARANTELLA inc. While concurrently Caldrea founded SCO Group, which was built on the purchased Santa Cruise Operations Group software.
They don't have a choice. Linux has already won. Heard of Android? Yeah that's the Linux kernel right there, over a billion mobile devices. Heard of Mac OS? That's a Unix kernel right there. Microsoft makes a proprietary operating system with cumbersome licensing restrictions. Microsoft is 10 years from death if it does not embrace Linux.
The IBM comparison is really apt. IBM hasn't died, they used to dominate like Microsoft, with all the same underhanded tactics, and they eventually came around.
There's also the "culture wars" to consider, where many (myself included) revered IBM for superior alternatives (OS/2), while old timers would just shake their head at anyone trusting them. Now, I shake my head at anyone trusting Microsoft, although they've tamped down their rabid attacks on open source and Linux, and they do offer some things that aren't complete garbage.
IBM at its peak was much more dominant than Microsoft ever was. Microsoft -- thanks to IBM (via PC DOS and Basic) -- eventually became the IBM of software. But Intel became the IBM of processors, Cisco the IBM of routers, Google the IBM of the web, and so on. IBM was the IBM of everything.
Incidentally, Microsoft co-developed OS/2 and there were only three problems with it: (1) you couldn't sell PCs running OS/2 -- IBM tried; (2) you couldn't get software houses to write programs for OS/2 -- except maybe Lotus; and (3) you couldn't get end users to install it. You had to install a whole OS to run OS/2, but Windows installed just like any other DOS program, and you could delete it if you decided you didn't like it.
Windows was selling well while OS/2 sales were, to quote Gates, "dismal", and Microsoft couldn't afford to support a failing project. (IBM may have had more programmers on OS/2 than Microsoft had programmers.)
Thanks to hiring Dave Cutler and staff from DEC, Microsoft could see a way out. However, IBM rejected the plan to move to NT as OS/2 3.0, and the two companies formally divorced. After that, IBM did its best to kill Microsoft. (One IBM senior manager told me: "We're going to burn Bill's butt.")
Bear in mind that in 1990, when Windows 3 came out, IBM had an annual turnover of $69 billion and Microsoft's turnover was $0.8 billion. It was a giant monopolist against a tiny start-up.
So, given the context, what was underhand about Microsoft's behavior, and what should it have done?
How did IBM's 1991 development of Workplace OS for PowerPC -- intended to cut both Microsoft and Intel out of the PC market -- fit into the strategy?
A period when IBM was spending more than $1 billion a year on OS/2 and WorkPlace OS to try to kill Microsoft? A period when IBM formed an alliance with Apple and Motorola to create a PC Reference Platform to try to kill both Microsoft and Intel? Up to the point in 1995 where IBM spent $3.5 billion on Lotus and used SmartSuite bundling to try to kill Microsoft Office?
Microsoft's turnover had grown to $5.3 billion in 1995, but IBM's turnover was $72 billion, and it was still a brutal monopolist targeting a much smaller company.
Indeed, people like me were cheering Microsoft back in that period.
I started programming in 1978, on IBM systems, and they were indeed "a brutal monopolist", probably feared more than Microsoft has ever been.
Sure, during the early-mid '90s Microsoft had sharp elbows and was looking out for its own interests, but those also just happened to coincide with ours. IBM was big, stupid and much more dangerous, UNIX(TM) was horribly fragmented (one of the core competencies I developed in the early '80s was porting software between different versions of it, and see The Unix Haters Handbook publishing in 1994: https://en.wikipedia.org/wiki/The_Unix-Haters_Handbook (disclaimer, Simson is a friend)).
GNU/the FSF was doing worthwhile stuff, but was and still is hanging fire on a kernel, the BSD porting efforts had been crippled by the '92-3 AT&T lawsuit, Linux was started from scratch in 1991 and immature, while NT was developed by well funded top notch experienced people (at least back then it was very rare if not unheard of for someone to develop more than one operating system, the process was so onerous) and was a great hope for a lot of us. In the mid '90s I developed or took part in developing several very solid systems using it, and I believe it's TCO and maintainability exceeded UNIX(TM) during that period.
Unfortunately, Microsoft started fumbling the ball about 3-4 years after NT's release (on or after the first Service Pack of NT 3.5 or more likely 3.51 as I recall), NT 4 was conceptually hopeless by moving graphics into the kernel, "and the rest is history".
Microsoft, of course, had a major problem, which was how to get the market to move from DOS-based Windows to NT. Not even offering OEMs 75% off the price of NT managed to shift them. Ultimately, Microsoft had to do whatever it took, and not all of it was nice. It didn't manage to do it until XP in 2001.
Re Unix, you have missed out the COSE/Hamilton group development of OSF/1, first released in 1991. This was supposed to unify most of the major players -- DEC, HP, IBM etc -- except Sun* behind a single Unix to prevent Microsoft from getting into corporate computing with NT.
Presumably you would have come across it....
* OSF = "Oppose Sun Forever" according to Scott McNealy
Yes, indeed, I used it in 1996-7 on the Alpha, just after it was renamed Digital Unix from DEC OSF/1 AXP. It was very nice, ran everything we wanted well, plus had a great High Availability capability. But I didn't view it as a major contender, back then at least, Sun provided the systems of choice. Assuming you could manage to buy them.
As for Win 16 -> Win 98 -> XP ... I don't know, I went from Windows 3.1 to NT, and as far as I could tell NT was better and more stable than the legacy non-memory protected stuff; hmmm, but I'll bet not for games (and NT4 putting graphics in the kernel was ugly). XP was great, and due to things mentioned elsewhere, I and then my parents used it for more than a decade.
Yes, NT was much more stable than Windows 95/98/98SE, and technically superior. Unfortunately it wouldn't run quite a lot of Windows programs, and one of them was whatever we used for AOL access, which was quite important at the time.
I dual-booted the two types of Windows way back then, and it was just too hard to stay on the NT side....
To some extent (possibly a large extent), OSF/1 was a push against the AT&T/Sun effort to unify Unix with SVR4. It just ended up more fractured than ever, which was a shame.
There was a brief window when Unix could perhaps have been more successful. That was between DOS and the Windows vs OS/2 war, when Microsoft Xenix was relatively successful. However, IBM was dead against Unix at the time, and OS/2 EE was meant to be the IBM-owned cornerstone of its Systems Application Architecture.
Unix could run on everything from PCs to Amdahl mainframes and supercomputers, whereas IBM had two major mainframe operating systems, at least five on minicomputers, plus workstations and desktops. No wonder IBM saw Unix as a massive threat to its account control.
If MS went with "NT OS/2" I doubt "OS/2 for PowerPC" would exist, which was a stripped down version of the "Workplace OS" you mentioned BTW. Interestingly, Windows NT themselves was ported to PowerPC too, and it was one of the main focus for NT 3.51 back in 1995.
Of course, the problems with this plan wasn't the WinNT part, it was the Win95/Chicago part. My favorite technical problem is how it's dependence on DOS led Caldera to continue their DR-DOS lawsuit. But it wasn't just the technical problems, there was also the tactics MS used to attack OS/2 2.x and later.
Win95 was cheap, easy to install (no reformat needed), had masses of software support, and sold well. Microsoft would have preferred to sell NT but the market voted with its wallet. (By the way, Win95 mainly used DOS as a loader.)
Otherwise, which tactics were unacceptable in the context of a small company being attacked by a giant monopolist that wanted to drive it out of business?
As I pointed out above: "there were only three problems with it: (1) you couldn't sell PCs running OS/2 -- IBM tried; (2) you couldn't get software houses to write programs for OS/2 -- except maybe Lotus; and (3) you couldn't get end users to install it. You had to install a whole OS to run OS/2, but Windows installed just like any other DOS program, and you could delete it if you decided you didn't like it."
Microsoft realized that you couldn't force people to buy something they (rightly or wrongly) didn't want, and decided to sell them something they did want. Small companies are like that. Being a global giant with a long history of monopoly power (going back to the 1930s), IBM didn't think like that.
I used to work with a guy who used to be the DP VP [what's called CIO these days] at the only Fortune 100 firm that didn't have a corporate IBM mainframe (although a subsidiary or two did, IIRC). IBM tried a number of times to get him fired (such tactics as golfing partners of the CEO saying that he was really incompetent, etc.). They played hardball, and probably still do, when they can.
Microsoft is not going to die if it doesn't embrace Linux... that said, they are much more open today than a decade and a half ago. .Net Core is making a lot of progress and has respawned interest in C# cross platform. I think a lot of this is in the light of what happened with Oracle v Google.
There's also, XBox itself and Azure, which are key points to the future of MS along with Office, which will now grow to other platforms, as it should have some time ago.
This assumes tomorrow MS doesn't drop a BSD licensed Android clone of their own and clean up in that market.
I thought Facebook was going to die out too then they became defacto telecoms in the developing world through WhatsApp/Internet.Org, and now have a billion users.
Yes, Linux has dominated the commodity server space. Look what happened to Sun & Solaris. But what about the desktop? Back in the late 90s there was much talk of how the Linux desktop would take over, KDE vs Gnome etc. IBM and Sun backed Open Office, Andy Herzfeld launched Nautilus. And then it didn't happen. MS still owns the business desktop with Windows & Office. And MS still has a big chunk of the consumer desktop, though Apple has taken a lot of it with MacOS.
People focused on innovating too quickly without building enough stability. Plus, programmers started imagining they could innovate as designers and we have Gnome 3, Unity, etc. All built for a mythical herd of grandmas who run linux, but need it to be user-friendly, in some imagined user-friendly way.
Plus, programmers started imagining they
could innovate as designers and we have
Gnome 3, Unity, etc.
I always thought those happened because someone allowed designers to make decisions on the projects? "Usability" I think has been the main driver behind a lot of the problems I have experienced over the last few years.
Yes for sure but the desktop itself is not the end of the game anymore, yes Microsoft still owns the desktop but on mobile and tablets, it's another story.
The desktop market is still mostly Windows & MacOS, but if we instead look at client devices in general, Microsoft is losing quite badly: Android (running a Linux kernel) and iOS (BSD-derived, like MacOS) have almost all of the market share, with Windows struggling to make much of a dent.
From my own perspective, both LibreOffice and Google Docs are both more then capable enough. I know that LibreOffice is primarily a desktop application, but it has a not-so-perfect remote web interface since version 3.5.
But, for filling out forms and keeping the design as it is, I'm currently using a web interface of Office 365. It's a bit annoying to use it at the moment (you can't get rid of the constant "Install Word on Your Device" messages and you first have to upload your document on OneDrive before you can edit it), but it gets the job done.
The problem is that other solutions are just not good enough to deliver the exact end product to send to clients or partners or management. We do everything in Google docs because it is just really nice and convenient but when it needs to be finalised for sending off we send it to the one person with a MS Office license who fixes it. I would never condone paying more licenses though because when we all had MS Office licenses exactly the same thing happened... People writing content are usually not the people doing layout and design.
Unless Microsoft ruins Office, I can't see it happening. I believe it would take a seriously funded company to implement all the needed features (sure, no one uses all the features of Word, but each uses a different vital set of them), prove that, oh, Excel was safe to run your company/trading on, and provide support.
People forget that MS used to have it's own Unix in the 80s: Xenix. I ported a large DEC PDP Fortran codebase to run on Xenix back then - it was far too big to fit in the 640K DOS limit. Writing the Xenix device driver for the 68000 powered graphics card in the PC was fun.
It's also very likely that if things went just a little differently, Microsoft's main OS today would probably be Xenix-based.
Microsoft created Xenix because they knew DOS would have a limited future, and in particular they knew there was a lot of value in running enterprise software on microcomputers. They also knew costs would go down and performance would go up over time and that running minicomputer-grade stuff on micros would become more and more feasible as time goes on.
They made Xenix for a number of microcomputer platforms, typically partnering with the hardware manufacturers (as is the case with Tandy). For the PC, they chose to partner with SCO instead for some reason.
Xenix was Microsoft's next-generation operating system, intended to be a platform for the few high-end customers who wanted to run enterprise software on PC, with plans to push it across the mid-range and even the low end as PCs get faster and fast hardware gets cheaper.
And then, it all changed when AT&T got broken up. Suddenly, AT&T was allowed to commercialize their software, including Unix. And AT&T was very, very much interested in doing this. The whole thing culminated in System V Release 4, but that took a few years for AT&T to put it together. In the meantime, Microsoft thought they'd never be able to compete in the Unix market against the creators of Unix, so they panicked, sold Xenix to SCO, and began looking for a partner for their _new_ next-generation operating system.
So Microsoft hooked up with IBM and created OS/2 as their successor to Xenix. Eventually, however, Microsoft fell out with IBM...
After falling out with IBM, Microsoft decided to do it entirely by themselves, so they poached a bunch of ex-VMS guys from DEC and created Windows NT. After a few years serving as an enterprise OS, Microsoft began trickling NT down to the consumer market, eventually replacing DOS-based Windows entirely with the launch of Windows XP in 2001.
Just imagine the alternate world where Microsoft didn't get spooked by AT&T commercializing Unix and instead focused on Xenix, trickled it down to the consumer market, and ultimately replaced DOS-based Windows with a Windows-style GUI built on top of Xenix.
IBM previously had told it's legion of "buying IBM won't get you fired" IT managers that the 286 powered PC-AT was the last PC model they'd have to buy for a long while (one reason Compaq was the first to come out a 386), and demanded OS/2 run on it.
Unfortunately, the 286 was seriously crippled for doing this sort of thing, as I recall, switching segment registers was very expensive. And for those of you who didn't live through the ugliness of 8086->286 (and 386 and beyond compatibility modes) programming, these were thoroughly 16 bit computers, where you had at most 64 KiB segments each of stack, code and data, unless you switched one of those segment registers to point elsewhere. That was cheap enough for the vast gain in memory on the 8086/8, and the 286 in its compatibility mode, but native? That mode did a lot more for you (like, hey, memory protection!), and Intel when that decision was made and implemented slowly simply didn't understand how people used their CPUs.
Anyway, Windows 3.0 and on was taking off, no one but stupidly managed Lotus wanted to program OS/2 Presentation Manager, and prior to in 1988 that DEC killed off a RISC CPU + new operating system effort in favor of continuing with VAX/VMS for a while longer. The software people, at least, were in Bellevue, WA, and many were hired to develop OS/2 3.0, which due to the above morphed into NT 3.1 (which really was around a 3.x Microsoft quality release :-).
(Seriously, it's a vile, vile hack, involving dynamically patching running binaries to work round Windows ABI differences and work around seg faults. But it works well enough to run a Debian userland.)
I have used Franz Lisp on VMS that had been built using Eunice. I guess the Eunice userland may have been installed somewhere on the VAX but I don't remember finding it.
Microsoft are now moving quickly to allow users to run .NET on all platforms. How long until they decide that Windows Server is no longer worth investing in?
Or are there too many Microsoft server products earning good profits (i.e. Exchange and BizTalk), to make that switch a realistic future expectation?
I'm genuinely interested to hear people's opinions on their thoughts of the future of Microsoft and Windows Server.
Heavy Windows server user here. We have 100+ windows server 2008 R2 and 2012 R2 machines in production and a massive .Net/C# codebase.
Currently it's a complete bastard of an operating system. It's expensive, hard to manage even with powershell and DSC etc, difficult to update, difficult to provision, complicated and to be honest absolutely terrible licensing hell that costs us days a year. Hyper-V just adds complexity before anyone suggests that.
Exchange on-site deployments are dead. Everyone is moving away from them now and into Office 365 and Google Apps. There is no rational cost justification to use anything else now. Even the big orgs (5000+ staff) are moving off it as it's cheaper to get a fat pipe in than it is to keep 2-3 windows admins and a pile of kit and a SAN on the payroll just to run groupware.
Biztalk is also dead and has been for a few years. People who were using it heavily seem to be holding onto it due to cost reasons and everyone else who has been using it and have done any platform re-engineering have moved off to Windows Server AppFabric, NServiceBus and custom integrations or to AWS/SQS.
Microsoft are pushing Windows Server 2016 with container support as the up and coming new thing. I'm really not interested in this myself. I don't think anyone else is either other than a select number of core windows bloggers. It's simply a "me too" move.
As we re-engineer our application, what our endgame currently looks like is deploying disposable Linux VMs (CentOS) on AWS running ASP.Net 5/golang, microservices, lots of small PostgreSQL nodes (RDS) rather than massive 48 core SQL Server boxes, using OAuth2/OpenID authentication and getting rid of our extensive operations team who are incidentally more of an obstruction than an aid to the organisation.
IMHO Windows Server is probably circling the drain. There are very few places it fits in a modern business and this is only going to get worse going forth as well.
The remaining killer is Active Directory but you can already get Microsoft to deal with that for you on Azure, pre-integrated with Office 365 and sharepoint etc.
Microsoft I suspect will end up a services company like Google with some hardware being sold on the side. And you know what? That's fine.
An even simpler summary: To add SPDY or HTTP/2 support to IIS, you must upgrade the operating system. I understand why this is (http.sys) but ... wow. Whereas using nginx I literally add "spdy" to the config and presto, double-digit% improvement on my site.
Yes exactly that sort of shit is what we have to deal with!
http.sys is a giant fucking turd. That and the whole urlacl thing. Nothing but pain. They added it because of shitty performance in the NT kernel when serving HTTP from userspace and it turned into a friction point for the whole operating system.
> Microsoft I suspect will end up a services company like Google with some hardware being sold on the side. And you know what? That's fine.
This seems to be one of the patterns for large enterprise software companies, IBM did it as well.
It would be interesting to see what Microsoft could do with Linux on the desktop if they swung their weight behind it, Microsoft has incredible engineering resources and some amazing programmers, I doubt it would happen but it would be fascinating.
Wow, I'm currently making nearly the exact same switch, albeit on a smaller scale. I'm currently in charge of a lot of old ASP.NET applications but for the new flagship thing we're looking at ASP.NET 5 on Linux on Azure (BizSpark), Go where it makes sense performance-wise or problem-wise and possibly Postgres instead of SQL Server, although I'm not sure on that last bit yet.
The plan for the original Active Directory was to be the directory service of a Microsoft dominated "information highway" called "The Microsoft Network" (v1). see the infamous Microsoft memo from Bill Gates
It seems they have big plans for their cloud product "Azure"
You're right in that Windows is complex operating system but how is it any different than CentOS? Sure, Powershell/DSC does make life easier but how are these utilities necessarily any worse or better than a combination of bash, python, kickstart?
Exchange Onsite deployments aren't dead, on-site email is dead. Unless, you're "too big" for the cloud (>20k users) because of <$Requirement> then Exchange is an option.
Keeping data On-Premise has always been a challenge, SAN are expensive but that's what it takes if you want data locally.
I can't comment on the rest but we're comparing Apples to Oranges here.
If you are struggling/fighting with the large deployments, I recommend you have a chat with http://www.devopsguys.com/ down in Cardiff, UK. That's all they do (and well). Speak to Jim and tell them @Junto sent you.
Thanks for the reference. They're not for this platform as we've already nailed down what we're doing but will keep them in my notes for future projects.
I know exactly what I'm talking about and am aware of the deprecation. I'm talking historically i.e. when people wanted managed long-running workflow hosting, monitoring and WCF correlation on the cheap instead of fishing out for Biztalk orchestration style integrations and licenses they jumped on that bandwagon.
"How long until they decide that Windows Server is no longer worth investing in ?"
Without hard data to support it, my impression (or better, my anecdotal experience) is that Windows Server has quite a strong hold in medium/large sized companies (not necessarily IT companies) as part of the local network infrastructure, and it's quite profitable for Microsoft.
I think it's treated as a necessary evil. People are starting to realise that it's less necessary these days.
Anecdotally, back a decade ago it was common to see a Windows 2000/2003/SBS box in even an SME environment being a fileserver/exchange/intranet/AD box. Now, I rarely see one. In fact they're declining even in large corps.
I'm currently in my first nix job after years with Windows. It's a larger company (15k+ people) and using RedHat for it's linux solution (solaris/aix is also present). Honestly, I don't understand the $$$ argument at the enterprise level against msoft. Not only are you paying Redhat for support but their RPMs are about a year behind what's out there (that's a lot of time!). I'm trying to get .net (coreclr) compiled now on RHEL to play around but having a heck of a time b/c of all the old packages (btw- if anyone has instructions for this, __PLEASE__ send along).
Are other nix shops using Ubuntu for their servers?
If you're comparing Red Hat RPM's version numbers, then don't. Red Hat works by backporting features and fixes on top of the old versions, precisely so customers don't need to upgrade to new major versions so often.
This always pops up when we have audits performed; Solutionary in particular seems to just go by version number with ssh, and doesn't seem to recognize that RHEL has backported numerous fixes into it's RPMs.
Looks like we'll be adding libunwind as a new package in RHEL 7.2. (There's a BZ, but for some reason I can't really fathom it is marked as private, and it's above my salary scale to open private bugs, sorry)
Lot's of Shops use Ubuntu, yes. However you could also use CentOS or Fedora to be more flexible then with RedHat, especially when it comes to costs.
'Mostly' you don't need the Enterprise support, however without any paying for it we won't have Fedora / CentOS in the future. ;)
As an hybrid kernel, able to do asynchronous IO better than open source UNIX clones, adoption of C++ instead of bare bones unsafe C and an OOP ABI (COM), I look forward to the existence of Windows as server OS.
There are lots of MS only shops out there and I don't see any of the changing in the near future.
When I first saw a whole stack of low latency C++ Win32 based servers running on Windows supporting an interest rate trading business back in 2004 I was amazed. Talarian SmartSockets was used for interprocess pub/sub between servers. There was no COM. No thread pools, just a handful of threads in each process executing distinct code. Very clean async design. Back then Linux didn't have lightweight threads, IIRC. I think this is the kind of thing Windows Nano might support. Is that what you had in mind when you look forward to Windows as a server OS?
> are probably only beaten by Solaris and Aix specific APIs.
They're definitely not. AIX copied the IOCP interface almost verbatim, but missed one critical piece: associating an IOCP with a desired concurrency level. Solaris added event ports, but again, they missed the concurrency link too.
The root of the problem on UNIX is the readiness I/O model.
You should see what you can do today with Registered I/O and the new threadpool APIs. It is phenomenal. (And literally decades ahead of anything UNIX offers. OS X comes close with GCD.)
Even though I used UNIX a lot on my career, I have always been with foots on both lands and enjoyed Win32 more than POSIX, claimed as portable, but full of nuaces per UNIX clone.
"C++ instead of bare bones unsafe C" is C++ really much safer than C?
"an OOP ABI (COM)" you really think this sort of thing is super great and there's nothing like it on "UNIX clones"?
I have heard interesting things about windows async i/o syscalls, but as you'll see in another sub-thread, IIS still has a major component in the kernel, because the overall syscall overhead of windows is apparently somewhat high.
This wouldn't be the first time I've heard of super-fancy windows subsystem designs which in practice were sabotaged by just generally over-complicated implementations. I recall a blog post by Mark Russinovich around the time Vista came out about the new, quite advanced file copying algorithms: http://blogs.technet.com/b/markrussinovich/archive/2008/02/0... - but that file copying performance was often significantly reduced when playing an mp3 file due to the "multimedia class scheduler" - http://superuser.com/questions/18600/107mb-s-network-file-co... - either because i/o and execution priorities didn't actually work, or the engineer assigned to do this didn't know what he was doing and neither did anyone he worked with. Who knows?
This sort of thing is all over microsoft's stuff. Theoretically super advanced, in practice just over-complicated. It's closed source and hugely complicated, so we'll never really know why it doesn't work as well as it should.
It still makes use of the original micro-kernel design,with kernel level RPC to separate modules. Thus still offering a micro-kernel cleanness in a monolithic space.
Also since Windows Vista many drivers are now implemented in user space, with minimal kernel primitives.
First of all, yes memory corruption will still bring the whole thing down. Only pure micro-kernels can survive it.
What Windows does, is that there aren't any direct calls between modules, only kernel level RPC calls.
All parameters are packed into the RPC message, it is dispatched, unpacked, validated as much as possible on the receiving end, then processed.
Messages and its validation introduce an indirection layer that improves safety.
It is based on the original micro-kernel design Windows NT had, before they moved graphics back into the kernel.
So the micro-kernel message passing is still there, but simplified since now everything runs on the same space. But as of Windows Vista, UDF was introduced for user space drivers.
The "Inside Windows" book series go into great detail how everything works.
imho, if Active Directory/ADAM ran on Linux demand for Windows Server would drop considerably. However Windows Server is also used to deliver Citrix-style remote desktops (using App Virtualization[1]) and that would be harder to replace. Also, SQL Server and Exchange use plenty of Windows-specific OS features -- it wouldn't be impossible to port to Linux but it would be a lot of work.
AD, Sharepoint, Office 365 is already canned in Azure/O365.
You won't be buying a Windows Server license to run them any more and it's feasible that they may even not run on Windows at some point in the future. Office's server portion and Sharepoint is all .Net so expect them to move it to ASP.Net 5 at some point.
SharePoint is a cash cow, but its implementation is such an old mess with lots of parts stayed unchanged since v2003. Nowadays Microsoft India is mainly working on it. The most polished version was v2007, since then little has changed.
SharePoint is basically the server part of Office. It's a web CMS, a web based Office (viewer), with AD account login support. I wonder why there is no open source SharePoint alternative e.g. coded in PHP.
>I wonder why there is no open source SharePoint alternative e.g. coded in PHP.
I think it's because SharePoint's strengths lie mainly in integration with Office and other Microsoft products. Lists, workflows, wikis, etc. in SharePoint aren't great. Basically everything it does can be done in a better way with another tool.
However, due to AD integration, Office integration, the ease of connecting to MS SQL Server, the product can do things that would be difficult to do in any third party product. You can upload Word documents, run a workflow on them, have users edit list data directly in Office products, bulk edit lists in Excel, and use InfoPath forms (basically dead now though from what I recall -- thankfully I no longer have to work with SharePoint).
I think SharePoint won't ever have a competitor that can offer all of its integration.
Actually Active Directory already ran on Linux. Samba4 has support for AD up to 2008 R2. Currently I don't think ADAM is builtin yet, but SAMBA4 is already in a good state, however it still misses a good GUI, (you need a MS OS to configure some stuff, like GPO)
Microsoft's foray - I don't necessarily mean that in a negative way - into OS technologies doesn't surprise me any more.
I remember an MS Rep telling us one day:
> Our CEO believes that consumption is the new currency.
The context in which he used it seemed to imply that Nadella believes that the way forward for Microsoft is to not so much compete with existing products, but to give people new and better ways to consume existing ones, and of course to consume these existing products themselves. So it really doesn't surprise me that they've built their own Linux for Azure Stack, especially with guys like Cumulus Networks in the market.
Sooner or later Microsoft will just build his GUI on Top of Linux and sell that or sell the support like Redhat. That would mean a lot to the computing World.
When .NET runs on more Devices they could just port more and more compatibility.
Microsoft's post revealing ACS says a fair bit about its features but doesn't explain why Microsoft felt decided to do with Linux distro? Perhaps the complexity of the world's switching ecosystem was the reason: Redmond says it has demonstrated ACS across with “four ASIC vendors (Mellanox, Broadcom, Cavium, and the Barefoot software switch), six implementations of SAI (Broadcom, Dell, Mellanox, Cavium, Barefoot, and Metaswitch)....
Per an Azure blog posting yesterday (http://azure.microsoft.com/en-us/blog/switch-abstraction-int...): "As of July 2015, the Switch Abstraction Interface (SAI) specification has been officially accepted by the Open Compute Project (OCP) as a standardized C API to program ASICs."
To my very limited understanding of this space, FPGAs and cheaper application-specific integrated circuits (ASICs) are essential to high speed networking.
It is a question of switch control more than anything. They are losing Cisco software, etc... with cheaper Open Compute switches, so they need some means to configure them within Azure; in all respects, that's a benefit to Microsoft. Facebook did the same thing: https://code.facebook.com/posts/843620439027582/facebook-ope...
Yes, the ASIC vendors provide drivers for Linux but not for Windows. MSR ported the Broadcom driver to Windows in the past, but it looks like Azure decided to take the path of least resistance.
Interix. Microsoft's Unix. It runs on the NT kernel alongside win32 as another process personality. Most windowses come with the Interix core, and you can get the userland part from Microsoft --- it's called Services For Unix these days, I think. It even comes with GCC. It's a pretty nice piece of work, behaving just like an old school Unix.
But it's largely been abandoned by Microsoft and is buggy as hell and difficult to install; and its an old old-school Unix, so it's missing a lot of the modern system calls and libraries that make Unixes nice. (I don't even know if win10 supports it.)
...that page cites one of my projects as a reference? May all the gods have mercy on our souls...
Yeah, it's obviously been destaffed at Microsoft. There's a blog, last updated years ago. http://blogs.msdn.com/b/sfu/
I know it's likely to be impossible, but if anyone at Microsoft's reading this (and chances are they are), whacking all the source code up on codeplex or github would be awesome. It worked so, so much better than Cygwin.
Not sure about Java, but Scala works fine for me in cygwin after doing this[1][2]. I'd have to have a more detailed example (that I can reproduce myself) of what gets messed up though if it's something beyond that, since I can't remember anymore exactly what else I had to do offhand.
Cygwin does not alter the behaviour of existing programs. It is a DLL that can be compiled into new programs to provide lots of POSIX/Linux compatibility. It also comes shipped with lots of pre-compiled Linux binaries of common tools.
If you run your pre-existing JRE, its behaviour won't change. If you run a JRE compiled with cygwin (I've no idea if they provide one or not) then it will have UNIX-style path support.
I'm not familiar with Mono or any of the .Net on Linux efforts, so forgive my ignorance. However, having a running .Net stack with a CLR VM and class libs is not the whole story when it comes to running Windows native server side stuff. One of the things I've discovered building spreadserve.com is that there's a lot of COM & Registry plumbing necessary to enable Windows native desktop binaries to run in a server environment. So I imagine running real .Net based servers on Linux would require something like Wine to supply ABI compatible interfaces to the registry and COM services, and lots else beside. That's a non trivial undertaking. The MS Linux described in the Register story sounds like it's dedicated to networking tasks. I'd bet there's no .Net running on it at all.
AFAIK Microsoft uses Mellanox Infiniband networking in Azure. So this Linux distro is most likely controlling Mellanox hardware. Do you know which Mellanox model of infiband switches they use?
More than just an OS, I think what they've built is a Network OS (NOS) on top of Linux. They have been involved in Switch Abstraction Interface(SAI) specifications from Open Compute Networking Projects from early days. I am excited what this brings on the table in Software Defined Networking. I really hope they open source their SAI implementation, and their NOS can be used with ONIE. So I could get a bare metal switch, install their NOS, and build network apps for my use-cases.
I used Windows for a little over a decade, then I discarded it one day when I suddenly realized that the proprietary OS actually never taught me anything about computers. Those precious 10 years went to waste. I'm ashamed to say that I couldn't discover Linux until about 5 years ago.
They're certainly refocusing but Azure and Office 365 are both growing rapidly. Of course, they're cannibalizing their own stuff. The interesting question is the effect on revenues a few years down the line....
Microsoft has also overtaken IBM in annual turnover, having given it a $50 billion/year start. That means it's refocusing from a healthier financial position, at the moment. It remains to be seen whether Nadella can do as well as Ballmer, who tripled turnover and doubled profits. That's quite a challenge.
It isn't necessary to "own" a growth market to be successful but having said that, don't discount Office 365 which has already usurped Google's similar offering by all accounts.
Lesson: Microsoft is willing to improve their proprietary offerings using leveraging free software others built. This is the main reason for FOSS uptake everywhere and makes great business sense for them. They also make a killing in patent royalties on the Android ecosystem. If anything, Nadella is focusing Microsoft more on the dollar than prior politics.
It's the same Microsoft, though, far as I can tell with many of the same tricks. All the spyware and schemes in their recent products confirms that. They shouldn't be trusted. People need to move away from their tech wherever possible. Except maybe the stuff like this that's built on tech someone can inspect, fork and/or clone.
On the other hand, that wasn't public like this. But it's a far cry from the two tries required to replace Hotmail's front end Apache FreeBSD servers with IIS? NT a decade or so earlier. Or the much more recent Danger Hiptop/Sidekick NetBSD + Java + sane servers -> Windows CE + Exchange that in probably the largest part doomed their $1 billion Kin effort: https://en.wikipedia.org/wiki/Microsoft_Kin which took only weeks to fail hard once it hit the market.
Holy crap, I didn't even know about the Skype event! Thanks for the link. Probably the best and funniest endorsement grsecurity ever had haha. However, I've enjoyed for years calling them on running Hotmail on FreeBSD and what it took them to replace the reliable AS/400 running their business with Windows servers. Funny shit.
Kin situation is another I didn't know about. Just wow. Most companies would have to work to screw up that bad. Microsoft is clearly a natural. Can't wait to see how much mockery (or praise) I have to offer at end of Nadella's reign. Honestly, though, I hope he transforms it operationally like Lipner's SDL did for security. Microsoft Research, in particular, has tons of potential that could inspire new stuff just to compete with whatever they offer. And then we buy the competing, hopefully-OSS stuff. ;)
Baller's consistent ability to create material losses no doubt had a large part in his downfall, see also e.g. the Xbox 360 hardware. And Intel is claiming that Windows desktop screwups are in part costing it a billion dollars in 1Q15, people aren't upgrading from XP.
Nadella I'm very uncertain about. To the extent he can terminate with extreme prejudice stack ranking and the poison it has suffused throughout Microsoft, he'll have much more people capital, including much less internal sabotage (part of why Kin failed). And he's completely changed how QA is done, although I don't know if for the better. On the other hand, he doesn't ... "feel" like the sort of person who can sufficiently directly change the culture for the better.
We'll see.
Ah, here's a bigger question: can Nedella change the perception of Microsoft's trustworthiness? That was bad under Gates, became catastrophic and frankly insane under Ballmer, and after so many have been burned by Microsoft has got to have an incalculable cost.
That last point is important. Even ignoring security, NSA, etc. there's the simple point of being a reliable partner. Microsoft has been almost random with how they're deciding on API's, desktop appearance, and so on. Then, without consistency from there. The amount they've fought with users over the damned start menu is probably the perfect case. The kludge I set up for a relative on Win 8 was ridiculous and whole experience reminded me why I'm on Linux (& Win7 previously).
Hopefully he turns that around. That alone might help a lot.
