Favorite Operating Systems of all time?

[zzip]

It's come a very long way, but after more than two decades of observation, I am inclined to say that it will be forever playing catch-up to systems with far more seats and thus have more engineering capacity thrown at them.

Of course. Windows defacto "owns" the Intel platform. So it will have the best support there.

 

On the other hand, there are other platforms that Linux "owns". Any device intended to run Android, a number of embedded platforms. Pi's and related devices. Presumably, Linux has the best hardware support on such devices.

[fujidude]

I was implying that the limited and tightly integrated scope of Macintosh hardware support is the reason that the out of box experience is well received.. However, as time went on, I found myself getting into situations where I needed LOTS of computing power in a portable package, which isn't something Apple wants to provide. So I've had to move into the elusive territory of mobile workstations (currently occupied by such beasts as the Dell Precision 7710, which I have, and the Lenovo Thinkpad P70/71 series.) these machines provided the computing power I needed (quad core xeon's running from 2-4GHz), and the ability to use ECC RAM (both are very important when you're running huge stacks of virtual machines to simulate data center deployments!) , these machines, while somewhat certified by Linux, do not have optimized drivers to fully utilize the hardware (Intel/NVIDIA Optimus hybrid GPU usage is very problematic, and the performance of the system with ONE 4K display was unacceptable, compared to the THREE 4K displays I run nominally on this unit.), and I was finding myself needing to use lots more software in the Windows domain, that didn't run well enough under Linux hosted VMs, so I had to switch to Windows 10.

 

As a side effect, I am able to run Linux very well in VMs, as well as having a Mac OS X guest that while not accelerated, does allow me to actually do software development in Xcode for iOS and MacOS app wanting clients, as needed.

 

What beast of a machine is this I am referring to? It's a:

 

Dell Precision 7710 Mobile Workstation

Intel Xeon E3-1505 4 cores 3GHz nominal peak, with near 4GHz possible with microcode tweaks.

64GB ECC RAM

4K IPS display

Intel P530 GPU and NVIDIA Quadro 5000M GPU with 8GB of DDR5 RAM

dual Samsung SM951 NVME SSDs (512GB running RAID-0)

one Samsung 850 PRO in the sata cradle (bringing total space to 1.5TB)

Fingerprint scanner and multi-format card reader for security

multiple USB 3.0 ports

USB-C for Thunderbolt 3 usage, attached to a Dell TB16 dock, providing my two displays, 6 additional USB 3.0 ports, and a USB-C (3.1 and TB3 and DP) port for monitor 0, and a DP port for monitor 1.

Two LG 27UD88 monitors, one in portrait, one (primary) in landscape

WASD keyboard (with custom font done in Harry Fat, to be very Atari-like, Atari fuji symbols replace windows symbols on keys)

3D Connexion cad-mouse

Shuttle Pro 2 controller for video work in LightWorks (I frequently have to make videos for clients often as demo or documentation)

 

-Thom

 

That's a hell of a "laptop." What does something like that set a person back? Also, what kind of work do you do with it?

[tschak909]

After discount, I got it for $5300. I paid for it with a foreign client job (you can write off foreign client work to a certain amount, and this happened to be under that amount).

 

I got it because I especially at the time was modelling out the deployment of data centers, by creating smaller stacks entirely in virtual machines running in OpenStack.

 

I needed a portable setup, because I need to bring the whole setup with me if I need to leave my lab and go to an office (thankfully, I do not have to do that often, my employers are often aware of my cerebral palsy).

 

Why the need to build a whole lab? Doesn't your employer provide lab space? Well, funny thing there.... With my previous employer, AT&T, we had a lab of 416 Dell R720xd's which were set aside for us to build big data infrastructure that could scale out. That didn't last long, as every fucking manager found out about our cache of machines, and muscled our managers to give them the computing space because "Gosh, they don't need all those, do they?" ... We literally saw our entire lab eroded because nobody wanted to pay for actual data center space, with redundant everything.... I'm sad now...

Whereas, with my current employer, virtually EVERYTHING that my team and I do, MUST be billable DIRECTLY to a customer. Let this sink in. This means, _NO LAB_, and _NO RESEARCH AND DEVELOPMENT ON IMPROVING WHAT WE CAN DELIVER TO CUSTOMERS._ .... I'm still sad.

 

To really bring this home, I was suddenly forced to move up the time table to purchase this system (I had spent the entirety of 2015 researching hardware to buy), and had narrowed it down to two vendors: Lenovo, with their P70, and Dell with the 7710. I _really_ wanted the Lenovo. It had faster CPU options, a better case design, better keyboard, and an additional drive tray that I could use to bring the total storage to 2.0TB. But, after talking with the IBM sales rep, they were not going to be ready to ship the system until March 2016. I had procured contract employment with my current employer on Nov 2015 (I became an employee Mar 2017), and less than three weeks on the job, our team had been tasked with BUILDING A COMPLETE EXAMPLE OF WHAT WE WERE GOING TO PROVIDE TO EMPLOYERS, and that we were to provide a DEMONSTRATION of it THREE DAYS BEFORE CHRISTMAS! (OR WE WERE GOING TO BE CANNED. NO JOKE.)

 

Lucky for me, I had just procured the laptop, and all seven of us, each given accounts and VNC and SSH connections logged into my beast of a laptop (I have symmetrical 150mb/150mb business FIOS to this house and the lab), I set up OpenStack, and we proceeded to BUILD A COMPLETE EXAMPLE OF A CONTINUOUS INTEGRATION PIPELINE spread across 12 virtual machines, and the laptop ran, continuously, for 3 weeks, everybody working 20 hour days, logging into it, and building it.

 

We made our demo. That beast of a laptop is the reason I have a job today, and therefore I consider it a complete fucking return on investment.

 

-Thom

Edited December 5, 2017 by tschak909

[TheTIGuy]

Would you quit ****ing cursing?

[tschak909]

No. I am truly sorry that you are so sheltered from the world that you feel that words should be censored, even when they convey honest emotional context..

 

-Thom

Edited December 5, 2017 by tschak909

[+OLD CS1]

You mean Sun *Cough* err... Oracle's Java SE Platform, right?

 

Well, Sun owned Java at the time but I mean Sony Ericsson's Java Platform (JP) developed for its phones. Hrmmmm I hope Oracle does not come looking for me now, but then if it does maybe I can finally buy a fekking support contract for my Solaris boxes.

[+OLD CS1]

Would you quit ****ing cursing?

 

Guy, lighten up. Nothing here we would not be saying sitting around a pub table.

 

No. I am truly sorry that you are so sheltered from the world that you feel that words should be censored, even when they convey honest emotional context..

 

-Thom

 

Well, at least he had the gumption to say something and not surreptitiously flag your message.

 

All that aside, looking at Linux versus Window, I have always been disappointed at how the Linux world never made desktop dominance (that and the never-realized resurgence and subsequent world take-over of the Amiga.) It seemed to me so much time was spent in-fighting, splintering, and so many distros trying to be Windows-alike. The argument I ran into often was "we have to be like Windows to get people away from it," and yet, and yet, and yet Apple Mac never tried to be like Windows.

 

In fact the opposite and admitted as such during the old TS2 seminar days. During the initial release days of Vista and the Office "ribbon" system we were told Microsoft was trying to adopt some of the Mac approaches. One of the guys in the audience quacks up with, "Who gives a fuck about Mac?" The presenter quickly shot back, "We do! When you go into the schools, into the labs, you see Apple on the desks, and when these kids get out into the world they go with and recommend what they know."

[tschak909]

Oh yeah, I tried to allude to the incessant in-fighting in one of my previous messages in this thread. I lived through it all, and watched the UNIX community not understand the reality of what was being fought for and where and why. We fought; and Microsoft watched, took notes, and went off and just did it. It's why we saw that massive intake of Windows NT in the mid 1990s, before Linux started fighting back in the server space.

 

There is always a time to deliberate, but the stability of a democratic system relies on the consensus of voting, if you don't do this, and just keep bickering, success becomes fleeting.

 

-Thom

[zzip]

 

Guy, lighten up. Nothing here we would not be saying sitting around a pub table.

 

 

Well, at least he had the gumption to say something and not surreptitiously flag your message.

 

All that aside, looking at Linux versus Window, I have always been disappointed at how the Linux world never made desktop dominance (that and the never-realized resurgence and subsequent world take-over of the Amiga.) It seemed to me so much time was spent in-fighting, splintering, and so many distros trying to be Windows-alike. The argument I ran into often was "we have to be like Windows to get people away from it," and yet, and yet, and yet Apple Mac never tried to be like Windows.

 

 

The "Linux Desktop" is like the Holy Grail. Linux conquered almost everything else, but it doesn't matter to them. "The Desktop, man! We need The Desktop. This will be the year, I can feel it!"

 

Look around though. In 2017 mobile is in many ways as important as the desktop was in the 90s, and Linux owns that though Android.

 

Truth is, there are only two ways that mass migration to a Linux desktop will happen. Either Linux gets a killer app that can't run on Windows or Mac, or Microsoft royally f***s up and makes living with Windows unbearable. And MS has done some pretty bad screw ups which has benefitted Linux in the past. But there is a large number of users who aren't adventurous enough to switch from what they know and will put up with quite a lot of MS-inflicted pain. And these days if they do switch, it will more likely to be to Mac. Mac has finally shown market share gains in the past five years, and they are far more common as corporate desktops than they used to be. And then at the other end of the market you have the rise of chromebooks.

 

So I don't think winning the desktop is important for Linux anymore. I think the 90s "fat desktop" metaphor is changing, and Linux should make sure its ahead of that curve and not worry about out-Windowing Windows.

[+x=usr(1536)]

We fought; and Microsoft watched, took notes, and went off and just did it. It's why we saw that massive intake of Windows NT in the mid 1990s, before Linux started fighting back in the server space.

Agreed, but there was one other significant factor that played into that: the death of VMS. Once NT was on the DEC Alpha platform, that entire end of the market had just been handed its gateway drug - and when Compaq started gradually killing off the Alpha platform in 1998, everyone who had moved to NT on Alpha had their stepping stone to NT on x86 by the time that Intel picked up the remaining pieces of the Alpha platform and eventually killed it off.

[tschak909]

True.

 

-Thom

[tschak909]

Even though I never saw how ANYONE sane could consider WNT 3.1 a replacement for VMS. Seriously. It was HORRIBLE.

 

-Thom

Edited December 6, 2017 by tschak909

[zzip]

Even though I never saw how ANYONE sane could consider WNT 3.1 a replacement for VMS. Seriously. It was HORRIBLE.

 

 

In my experience it was management types who drank the Kool-Aid but never got their hands dirty with either OS. I kept hearing "NT is the future" refrain in the late 90s, but kept thinking to myself "how can this crap be the future?". It was around that time I really became a strong Linux advocate. Apparently I wasn't alone because the entire open-source scene exploded and started going mainstream soon after

[+x=usr(1536)]

Even though I never saw how ANYONE sane could consider WNT 3.1 a replacement for VMS. Seriously. It was HORRIBLE.

 

-Thom

Or even NT4 pre-SP4. I remember having to work with machines in that config, and about the best you could say for them was that at least they weren't NT3.x.

[+x=usr(1536)]

In my experience it was management types who drank the Kool-Aid but never got their hands dirty with either OS.

A lot of what I saw was that, to be sure, but also fear of deploying an OS with no contracted support behind it. More on that further down.

 

I kept hearing "NT is the future" refrain in the late 90s, but kept thinking to myself "how can this crap be the future?". It was around that time I really became a strong Linux advocate.

Yeah, but remember that at the time Windows 3.1 had been out for a few years and become the de facto networked office desktop environment. NT was a way to provide services for environments like that in a relatively integrated manner while also opening up the possibility of a workstation OS within the same ecosystem for people who needed more grunt than Win3.1 could provide.

 

That was the promise, anyway. I remember those networks, and NT on x86 was a crashy POS. Pretty good on Alpha, but utter crap on x86 until SP4 floated.

 

Apparently I wasn't alone because the entire open-source scene exploded and started going mainstream soon after

RedHat were the ones who really cracked it: by providing contractual support for the OS, they removed a lot of the concern of running an open-source OS in enterprise. Sure, they weren't the first to come up with the support model, but they were the ones who really figured it out.

[zzip]

A lot of what I saw was that, to be sure, but also fear of deploying an OS with no contracted support behind it. More on that further down.

 

Yeah, but remember that at the time Windows 3.1 had been out for a few years and become the de facto networked office desktop environment. NT was a way to provide services for environments like that in a relatively integrated manner while also opening up the possibility of a workstation OS within the same ecosystem for people who needed more grunt than Win3.1 could provide.

And that makes sense. But at the time I was working on applications running on HP/UX and I kept hearing that NT was going to replace Unix, but NT seemed ill-designed for the type of applications I had. NT was still rather desktop-oriented, and so were its applications. It was clunky to run as a "headless" system. Often times you had to physically visit the server to click a button on a dialog-- stuff like that. You couldn't remotely access it from home in the days of dial-up.

 

RedHat were the ones who really cracked it: by providing contractual support for the OS, they removed a lot of the concern of running an open-source OS in enterprise. Sure, they weren't the first to come up with the support model, but they were the ones who really figured it out.

True about Redhat. But what the floodgates really seemed to open when Netscape announced it was going open source with its browser in 98. After that, corporate interest in open source seemed to rise rapidly.

[+OLD CS1]

RedHat were the ones who really cracked it: by providing contractual support for the OS, they removed a lot of the concern of running an open-source OS in enterprise. Sure, they weren't the first to come up with the support model, but they were the ones who really figured it out.

 

Yes and no. I supported a customer running RHEL 5 for web hosting and he had a support contract with Red Hat. He put together a system based on the HCL but when we had problems with the RAID controller for which RHEL 5 had the driver built-in, Red Hat support told me the vendor had to help me because they cannot support hardware or the drivers, then the vendor told me that was horse-puck and the whole reason they submit the driver to Red Hat is to be validated and then supported by Red Hat.

 

Needless to say we got nowhere and his RAID controller kept eating his array.

 

There is always a time to deliberate, but the stability of a democratic system relies on the consensus of voting, if you don't do this, and just keep bickering, success becomes fleeting.

 

I would like to hear your opinion on how Linus runs the Linux project. I believe he has a management role which at times must over-rule the democratic process. I also know there are a lot of programmers out there who are very good and very sure of themselves and do not like to be managed even though the necessity exists.

[tschak909]

Linus has always been described as a "benevolent dictator" .. and, while I don't agree with that personality assessment (he is highly opinionated, and can easily have his feathers ruffled if you step into his domain of expertise, thing is the kernel is so large that this doesn't happen as often in practice...), he does run an organizational structure that has worked and has consistently produced release after release for 26 years, now.

 

His lieutenants have done a very good job with a singular goal to produce a kernel that builds and works for the majority of people, and this is a big part of why Linux as a project is successful, as it became more and more popular, delegation and carefully built trusts between the lieutenants and the various developers they interact with have produced a mostly stable ecosystem, if very volatile at times. The trust is such that if Linus has to step in and make a final decision, a rough consensus often accepts it, and for those that don't, there's a fork.

 

So, while it is not completely democratic in terms of what goes into the main line kernel, the ability to fork, and/or keep submitting refined patches until acceptance does make it a far more equitable process than others. Linus is all about code, and I've seen patches come from far flung places, if there's a rough consensus between Linus and his Lieutenants, there's a good chance the patch will be accepted, and patches continue to be accepted at a high rate, whether they come from individual or company.

 

Has Linus thrown fits? PLENTY. Have his Lieutenants thrown fits? PLENTY.

 

But through it all, the project has remained stable, and I commend them for that.

 

-Thom

[+x=usr(1536)]

Yes and no. I supported a customer running RHEL 5 for web hosting and he had a support contract with Red Hat. He put together a system based on the HCL but when we had problems with the RAID controller for which RHEL 5 had the driver built-in, Red Hat support told me the vendor had to help me because they cannot support hardware or the drivers, then the vendor told me that was horse-puck and the whole reason they submit the driver to Red Hat is to be validated and then supported by Red Hat.

 

Needless to say we got nowhere and his RAID controller kept eating his array.

Yeah, you've got my sympathies - I've run into the same thing. To be fair, though, that doesn't strike me as a problem unique to either that vendor or RedHat; I've seen the stuff like this happen on other platforms under other similar support contracts.

 

By any chance did the hardware vendor's name rhyme with 'hell'? This sounds remarkably similar to a problem I had a few years after yours.

 

Ordered six servers; fBSD 6.x was the target OS for them. HCL from the vendor and fBSD project claimed that the onboard Broadcom gig-E adapters were 100% compatible. This turned out to be true... On two of the six boxes. Neither the hardware vendor nor anyone assisting on the fBSD side could figure it out until we noticed that, at some point during hardware production, the NICs had been changed to a slightly different part number but were still marked the same as the old ones and classed as being in the same family. Of course, the drivers didn't work.

 

That one I do pin on the vendor, and we were dealing with community support since, well, it was fBSD... But much of the same fingerpointing ensued.

[+x=usr(1536)]

And that makes sense. But at the time I was working on applications running on HP/UX and I kept hearing that NT was going to replace Unix, but NT seemed ill-designed for the type of applications I had. NT was still rather desktop-oriented, and so were its applications. It was clunky to run as a "headless" system. Often times you had to physically visit the server to click a button on a dialog-- stuff like that. You couldn't remotely access it from home in the days of dial-up.

Understood. And I can say that it was possible to remotely access an NT4 box with GUI over dialup - provided you were running (IIRC) RAS on Terminal Server edition and really, Really, REALLY liked the remote access equivalent of repeatedly punching yourself in the nuts.

 

I am *so* glad those days are over.

Edited December 6, 2017 by x=usr(1536)

[+OLD CS1]

Yeah, you've got my sympathies - I've run into the same thing. To be fair, though, that doesn't strike me as a problem unique to either that vendor or RedHat; I've seen the stuff like this happen on other platforms under other similar support contracts.

 

By any chance did the hardware vendor's name rhyme with 'hell'? This sounds remarkably similar to a problem I had a few years after yours.

 

Ordered six servers; fBSD 6.x was the target OS for them. HCL from the vendor and fBSD project claimed that the onboard Broadcom gig-E adapters were 100% compatible. This turned out to be true... On two of the six boxes. Neither the hardware vendor nor anyone assisting on the fBSD side could figure it out until we noticed that, at some point during hardware production, the NICs had been changed to a slightly different part number but were still marked the same as the old ones and classed as being in the same family. Of course, the drivers didn't work.

 

That one I do pin on the vendor, and we were dealing with community support since, well, it was fBSD... But much of the same fingerpointing ensued.

 

Yeah, been there, too! I used to keep a bunch of Intel 82550 and 3Com 3c905 cards on-hand simply because we ran into so many issued with different network cards, on Linux, Solaris, and Windows NT/2000. No matter what, these cards ALWAYS worked.

 

No, it was not Dell, it was American Megatrends (AMI.) You are definitely correct this is something which is not limited to Red Hat, but the problem is you would expect a different level of support to follow the hype.

 

Understood. And I can say that it was possible to remotely access an NT4 box with GUI over dialup - provided you were running (IIRC) RAS on Terminal Server edition and really, Really, REALLY liked the remote access equivalent of repeatedly punching yourself in the nuts.

 

Oh, how I long for the days.

[tschak909]

AMI are the original vendors of the MegaRAID, Dell licensed it for a short time, before building the PERC's in-house.

 

-Thom

[+OLD CS1]

AMI are the original vendors of the MegaRAID, Dell licensed it for a short time, before building the PERC's in-house.

 

MegaRAID were really damn good controllers, IME. Ran them under Solaris 8 x86 and Windows. But I am really going OT, now

[zzip]

A lot of what I saw was that, to be sure, but also fear of deploying an OS with no contracted support behind it. More on that further down.

It wasn't that I expected people to replace NT with Linux at the time. It seemed like a hacker OS that would have niche corporate acceptance at best, but there were a number of Unix vendors that provided those kinds of support contracts for their own platforms. Many of them didn't seem to put up much of a fight against NT creeping into their space. It seemed like this was what caused people to rally around Linux and make it acceptable as an NT-alternative.

[Flojomojo]

I used NT for work and appreciated the stability, but wouldn't use it at home until it supported some games. That's where Windows 2000, Me, and XP were great.

 

We are all coming at this from different directions, with wildly varying expectations for usefulness, cost, and fun out of an operating system, all informed by our individual experiences. Opinions are great!