Hmm... comment screwed up, should have used preview...
My first statement was along the lines of: you can get an extra 256mb for less than $50 au. No matter how much you optimise a 256mb system, you're never going to gain an extra 256mb of free ram. Is the time spent worth less than $50 to you?
Extra 256mb of RAM
I mean, don't get me wrong, i'm all for conserving memory as appropriate, but really is the cost in convenience worth less than $50 (i'm talking $AU - this will get 2 people to the movies and popcorn + coke over here) to you?
Back in the day when I was trying to run X in 8mb on my 486 in 1995, and upgrading to 16 meg would have cost me about $400 or more, I was optimisation crazy, and you really could make some good gains. There comes a point of diminishing returns though...
There are docs for DSL linux that tell you how to do it - it takes a little buggering around.
However, I've run into issues when trying to boot on SATA systems, as it treats your USB drive as a SATA device, and the root device listed in your usb stick then refers to the SATA disk in the machine.
I'm pretty sure you could work around that by specifying boot parameters when booting (for the correct root device) but haven't looked into it much.
Given that the youth of America have been brought up on MS products, they're going to have a stronger attachement to them than those of us who were brought up on Commodores, Amigas, and Apples. MS *clearly* knows this. Think about that.
Yes and no. Kids are curious. Kids like to try and go against the grain, find their own little niche, to be "cool" or "different".
It would not surprise me in the slightest if we have a lot of them experimenting with Linux and other alternative OSes on that basis alone. I know back when I started playing with Linux when I was 17-18 in the mid 90s it was initially largely because it was obscure and different and i was curious about it:D
For any machine that doesn't need sound or graphics and only rarely changes hardware, Linux kicks serious ass. For the rest, I hope you have the exact same rev of the exact same hardware and run the same version of the same distro as someone who wrote a HowTo article, or get ready for some pain.
On the contrary, I've changed motherboards quite happily several times before with Linux without missing a beat or having to reinstall (desktop machine, and failed firewall/router boxes).
Usually, if you're running a semi-sane kernel without everything stripped out (and in the days of multi-gigabytes of RAM, who cares about a couple of extra meg in kernel size) usually the worst case scenario is boot from recovery disk, modify your boot loader and job done - 10-15 minutes work, tops.
Try doing that with XP.
Documentation issues I'll give you - this is why i generally choose FreeBSD instead. FreeBSD may sometimes lag in features slightly (though personally I haven't had any un-workable issues with it), but generally once features are implemented, the interface/API doesn't change simply for the sake of change.
Not sure I really agree with that. My first exposure to the net was on a Mac running Netscape 1.0.
Sure, Windows 95 was an option for Internet use, but if I really don't think it was the catalyst for making it popular. The number of Mac and Windows 3.1 customers I dealt with in the 90s indicates that even if Windows 95 was never released, the internet would have taken off anyway. What really accelerated growth in my opinion was the graphical web browser and decent speed modems. And for that we have to thank Mosaic...
This is exactly what you get if Linux or any other free O/S deletes your data for no apparent reason. HOWEVER if you read the EULA for Windows, it is ALSO exactly what you get if Windows does the same thing.
If you can't understand that "8. Account Inactivity. Lycos reserves the right, in its sole discretion, to delete any materials (including emails) stored in connection with an unpaid Lycos Mail account if the user's account has been inactive for thirty (30) days." means that Lycos may delete your data at their leisure if you don't log in, then no amount of customer service will help you.
Erm... "the internet" originated on unix boxes. If by "the internet" you mean "teh intarweb" then that originated on NeXT. Microsoft was quite late in getting on the internet bandwagon - it wasn't until Windows 95 that they even shipped a TCP/IP stack...
Sorry, kids, but although I can deal with running a Gentoo system, I choose to run Kubuntu 6.10. Not because I'm too much of a wuss to run Gentoo, or because I'm too stupid to run anything other than Ubuntu, but because I'd rather spend the hour or so of computer time I have at home some days getting pix and video of my adorable girl (now at toddler age) ready for the grandparents.
What he said.
Myself, I started off with Slackware, and yes, it's neat to be able to compile all your own software, etc - but eventually you come to realise that all you're doing is increasing the uniqueness of your particular machine, which in the scheme of things is not really wise. If you have a problem, it's going to fall back to you to fix it, possibly without a lot of help and/or false advice (from well-meaning people who may accidentally mis-diagnose your problem) along the way.
I think all of us go through the "this is neat!" stage, and then eventually after several years you realise that you're just creating more work for yourself, and it's time that could be better spent doing something else (be it work related, open-source contributions or more relaxation time - whatever).
After playing these games (*nix system administration) for 12+ years, my advice is to keep your platform as "standard" as possible and only recompile stuff with "special" optimisation or use flags if there's a demonstrated *need* to do so for that particular application. And even then, most of the time any performance improvements are usually in the region of 10-20% or less, at the cost of software commonality with the rest of the world.
Compared to downtime or increasing complexity (and hence, maintenance costs), hardware is (usually - unless you're at the extreme high end) cheap:)
This kind of thinking made sense in 1995, when your servers were only hooked up to the office LAN. But now it's 2007 and they are hooked into a global network, on which millions of nefarious people are working hard every day to find new ways to crack systems.
This makes the "if it works, don't update it" line of thinking not just misguided but actively dangerous, because it means that you're not keeping up with security fixes that are issued to close avenues of attack that were never envisioned when the product first shipped. (Unless those fixes are included in the author's definition of "new features", that is. But I kind of doubt that.)
Erm, i don't think the poster was meaning to skip security updates. An insecure machine is not "working just fine" in my book, and security updates are a seperate issue to chasing application versions.
You get it working and *leave the system alone* except for security fixes. Any halfway decent server O/S will provide you with the option for security-only updates that do not make changes to functionality at all.
I've seen it a number of times - many system admins recommend wiping and reinstalling Windows on a yearly basis as it clogs up with various installers, viruses, adware, spyware, basic junk, etc.
It's not required to reinstall Windows yearly or bi-annually or whatever; you just need to be extremely disciplined in what you allow on to your system.
However, I do agree with your major point: requiring the presence of an old O/S on the disc is retarded. When you upgrade SCSI controller or whatever and XP can't install to it due to lack of drivers because it's considered end of life, you're screwed.
In a production environment, I can see zero reasons to go with Gentoo. Whilst i'm sure there's going to be plenty of people claiming "if you can't use it you shouldn't be a sysadmin" or "it's so optimised" or "use flags rock!!", etc - all of those things are missing the point.
On a server that needs to be stable, in a production environment, you want to minimise the possible variables to deal with, minimise your exposure time to known threats and minimise the possibilities for making mistakes.
By compiling your own binaries, sure - if you're really anal you can examine the source to each and every program to make sure it's not trojan-ed. Sure, you can tweak use flags and optimisation flags as much as you like. However, if you do that you're running binaries that are probably quite different to the majority of the rest of the gentoo population.
If something goes pear shaped, who's problem is it? Is it an issue with your compiler? Your hardware you used to compile with? Your use flags? Your optimisation flags? Some obscure library out of date? Who knows? And when it's broken and down for a few hours, your boss' opinion is likely to be "who cares? just fix it".
If you're running a "known good" binary that has been compiled with "known good" flags, you minimise all that.
Sure, it won't be optimised, but 99% of the time with today's hardware, it doesn't matter a shit. And *if* it does, locate the bottleneck/hotspot and optimise *that* package only so you can keep the rest of your system as close to "standard" as possible, and you'll reduce your exposure to "wierd" problems that most other people don't encounter.
I don't have any issues with Gentoo as a distribution, but seriously, horses for courses... I wouldn't personally choose to run OpenBSD as a multimedia desktop. By the same token, I wouldn't suggest running bleeding edge roll-your-own gentoo (roll-your-own-compile-flags FreeBSD for that matter) on a server, either.
People are demanding interactive pages that Sally the GM's personal assistant can update as required. "Web masters" are no longer really called web masters, they've been replaced with graphic designers and application developers...
So what do they do when the next critically bad exploit comes out for Win2k, which is by now well past end of life?:)
I like win2k over xp as much as anyone, but there comes a point where if it's either your ass or microsoft's, you bite the bullet and make sure it's microsoft's:D
Gamers: PS3 and Wii, and XP (no game co's will make for one OS only)
Maybe not for the first 6-12 months, but if you think that no game company is going to embrace directX 10, you are mistaken. How many current directX 9.0c only games are there? Like... most on the shelves released within the past 12 months...
Media users? You mean the ones who buy shit on iTunes? They're going to use Linux? Right....
Regular companies? Running Win2k? Maybe those with less than 30 employees - any bigger than that and they're going to be running Vista by the time XP is end of life.
Why backup operating systems and consume precious, expensive backup space?
Whilst I agree that LTO autoloaders, etc are expensive... in the scheme of things, backing up the O/S is a piss in the ocean. For example, we back up 2.3tb of data per week with daily differentials. Allowing say 1gb for each server o/s back up, we're looking at say 15gb of O/S recovery data.
It doesn't even register... even on a single 400gb LTO3 tape, its insignificant.
Will the Windows Vista content protection board robustness recommendations increase the cost of graphics cards and reduce the number of build options?
Everything was moving to be integrated on the one chip anyway and this is independent of content protection recommendations. Given that cost (particularly chip cost) is most heavily influenced by volume, it is actually better to avoid making things optional through the use of external chips. It is a happy side effect that this technology trend also reduces the number of vulnerable tracks on the board.
Oh happy joy! Clearly microsoft is in bed with RIAA/MPAA. Now, i'm a firm believer that most of the stuff they (RIAA/MPAA) is shit, and I don't watch/buy/download it, but the thing is, if i buy a movie, or what have you, i licensed the content. I should (morally) be able to do whatever I like with said content for my own personal use, be it convert to another device, or whatever (classed as "fair use"). DRM technology eliminates that and requires that whenever a platform shift happens i need to re-buy my content.
How the fuck is technology that encourages/mandates this a "happy side effect"?
Slashdot Mirror
Maybe apple should use their monopoly on portable media players to leverage OS/X, and not bother to write any vista iTunes software at all :D
My first statement was along the lines of: you can get an extra 256mb for less than $50 au. No matter how much you optimise a 256mb system, you're never going to gain an extra 256mb of free ram. Is the time spent worth less than $50 to you?
Back in the day when I was trying to run X in 8mb on my 486 in 1995, and upgrading to 16 meg would have cost me about $400 or more, I was optimisation crazy, and you really could make some good gains. There comes a point of diminishing returns though...
I think the old saying that "People who like unix use BSD, people who hate Microsoft use Linux" applies here more than ever :)
Thanks for trying...
However, I've run into issues when trying to boot on SATA systems, as it treats your USB drive as a SATA device, and the root device listed in your usb stick then refers to the SATA disk in the machine.
I'm pretty sure you could work around that by specifying boot parameters when booting (for the correct root device) but haven't looked into it much.
Yes and no. Kids are curious. Kids like to try and go against the grain, find their own little niche, to be "cool" or "different".
It would not surprise me in the slightest if we have a lot of them experimenting with Linux and other alternative OSes on that basis alone. I know back when I started playing with Linux when I was 17-18 in the mid 90s it was initially largely because it was obscure and different and i was curious about it :D
On the contrary, I've changed motherboards quite happily several times before with Linux without missing a beat or having to reinstall (desktop machine, and failed firewall/router boxes).
Usually, if you're running a semi-sane kernel without everything stripped out (and in the days of multi-gigabytes of RAM, who cares about a couple of extra meg in kernel size) usually the worst case scenario is boot from recovery disk, modify your boot loader and job done - 10-15 minutes work, tops.
Try doing that with XP.
Documentation issues I'll give you - this is why i generally choose FreeBSD instead. FreeBSD may sometimes lag in features slightly (though personally I haven't had any un-workable issues with it), but generally once features are implemented, the interface/API doesn't change simply for the sake of change.
Sure, Windows 95 was an option for Internet use, but if I really don't think it was the catalyst for making it popular. The number of Mac and Windows 3.1 customers I dealt with in the 90s indicates that even if Windows 95 was never released, the internet would have taken off anyway. What really accelerated growth in my opinion was the graphical web browser and decent speed modems. And for that we have to thank Mosaic...
This is exactly what you get if Linux or any other free O/S deletes your data for no apparent reason. HOWEVER if you read the EULA for Windows, it is ALSO exactly what you get if Windows does the same thing.
Perhaps you would like a refund?
Erm... "the internet" originated on unix boxes. If by "the internet" you mean "teh intarweb" then that originated on NeXT. Microsoft was quite late in getting on the internet bandwagon - it wasn't until Windows 95 that they even shipped a TCP/IP stack...
Apparently it "starts your wow" whatever that means :D
Dude, they've been working on WinFS since about 1993. If it's *still* not ready yet, it never will be...
I wouldn't really suggest ubuntu on a serious production server, either... desktop, no problems tho...
What he said.
Myself, I started off with Slackware, and yes, it's neat to be able to compile all your own software, etc - but eventually you come to realise that all you're doing is increasing the uniqueness of your particular machine, which in the scheme of things is not really wise. If you have a problem, it's going to fall back to you to fix it, possibly without a lot of help and/or false advice (from well-meaning people who may accidentally mis-diagnose your problem) along the way.
I think all of us go through the "this is neat!" stage, and then eventually after several years you realise that you're just creating more work for yourself, and it's time that could be better spent doing something else (be it work related, open-source contributions or more relaxation time - whatever).
After playing these games (*nix system administration) for 12+ years, my advice is to keep your platform as "standard" as possible and only recompile stuff with "special" optimisation or use flags if there's a demonstrated *need* to do so for that particular application. And even then, most of the time any performance improvements are usually in the region of 10-20% or less, at the cost of software commonality with the rest of the world.
Compared to downtime or increasing complexity (and hence, maintenance costs), hardware is (usually - unless you're at the extreme high end) cheap :)
Erm, i don't think the poster was meaning to skip security updates. An insecure machine is not "working just fine" in my book, and security updates are a seperate issue to chasing application versions.
You get it working and *leave the system alone* except for security fixes. Any halfway decent server O/S will provide you with the option for security-only updates that do not make changes to functionality at all.
It's not required to reinstall Windows yearly or bi-annually or whatever; you just need to be extremely disciplined in what you allow on to your system.
However, I do agree with your major point: requiring the presence of an old O/S on the disc is retarded. When you upgrade SCSI controller or whatever and XP can't install to it due to lack of drivers because it's considered end of life, you're screwed.
On a server that needs to be stable, in a production environment, you want to minimise the possible variables to deal with, minimise your exposure time to known threats and minimise the possibilities for making mistakes.
By compiling your own binaries, sure - if you're really anal you can examine the source to each and every program to make sure it's not trojan-ed. Sure, you can tweak use flags and optimisation flags as much as you like. However, if you do that you're running binaries that are probably quite different to the majority of the rest of the gentoo population.
If something goes pear shaped, who's problem is it? Is it an issue with your compiler? Your hardware you used to compile with? Your use flags? Your optimisation flags? Some obscure library out of date? Who knows? And when it's broken and down for a few hours, your boss' opinion is likely to be "who cares? just fix it".
If you're running a "known good" binary that has been compiled with "known good" flags, you minimise all that.
Sure, it won't be optimised, but 99% of the time with today's hardware, it doesn't matter a shit. And *if* it does, locate the bottleneck/hotspot and optimise *that* package only so you can keep the rest of your system as close to "standard" as possible, and you'll reduce your exposure to "wierd" problems that most other people don't encounter.
I don't have any issues with Gentoo as a distribution, but seriously, horses for courses... I wouldn't personally choose to run OpenBSD as a multimedia desktop. By the same token, I wouldn't suggest running bleeding edge roll-your-own gentoo (roll-your-own-compile-flags FreeBSD for that matter) on a server, either.
People are demanding interactive pages that Sally the GM's personal assistant can update as required. "Web masters" are no longer really called web masters, they've been replaced with graphic designers and application developers...
I like win2k over xp as much as anyone, but there comes a point where if it's either your ass or microsoft's, you bite the bullet and make sure it's microsoft's :D
Maybe not for the first 6-12 months, but if you think that no game company is going to embrace directX 10, you are mistaken. How many current directX 9.0c only games are there? Like... most on the shelves released within the past 12 months...
Media users? You mean the ones who buy shit on iTunes? They're going to use Linux? Right....
Regular companies? Running Win2k? Maybe those with less than 30 employees - any bigger than that and they're going to be running Vista by the time XP is end of life.
Whilst I agree that LTO autoloaders, etc are expensive... in the scheme of things, backing up the O/S is a piss in the ocean. For example, we back up 2.3tb of data per week with daily differentials. Allowing say 1gb for each server o/s back up, we're looking at say 15gb of O/S recovery data.
It doesn't even register... even on a single 400gb LTO3 tape, its insignificant.
Oh happy joy! Clearly microsoft is in bed with RIAA/MPAA. Now, i'm a firm believer that most of the stuff they (RIAA/MPAA) is shit, and I don't watch/buy/download it, but the thing is, if i buy a movie, or what have you, i licensed the content. I should (morally) be able to do whatever I like with said content for my own personal use, be it convert to another device, or whatever (classed as "fair use"). DRM technology eliminates that and requires that whenever a platform shift happens i need to re-buy my content.
How the fuck is technology that encourages/mandates this a "happy side effect"?
Might i point out that my dreamcast (that I actually purchased *after* my PS2) is broken. My PS2 is not.