That actually IS part of the stupid "get the facts straight" campaign. What this and other silly comparisons always conveniently leave out is the fact that a good Linux admin can handle 50-100 boxes (more in an identical cluster situation), whereas a good windows admin can handle 20-30 boxes.
How about some math: (these salaries are made up and meant to illustrate that even if your Linux admin makes DOUBLE what your Windows admins make, it's still a better deal.)
Linux admin: $80K, Win admin: $40K
30 machines: Linux $80K, Win $40K 30-100: Linux $80K, Win $160K 100-200: Linux $160K, Win: $280K
So now let's try it with closer to real figures: Linux admin: $50K, Win admin: $45K 30 boxes: Linux $50K, Win $45K 30-100: Linux $50K, Win $180K 100-200: Linux $100K, Win $315
If Windows admins are in fact paid less, Windows can only come out ahead in small operations. Furthermore, the closer Win and Linux admin salaries are, the better the savings are in a large operation when you go with Linux.
Also as an admin, if you see that even Microsoft is saying that Linux admins make more money, why would you waste your time learning their system? Show me the MONEY!
I request that any word documents sent to me via email are reformatted as plain text, RTF, or PDF, with my preference being plain text.
I delete any incoming Word docs as a precautionary measure; I don't feel like there's any supporting evidence why I _should_ accept Word docs. If the user can't figure out how to reformat the document, I feel that there's a damn good chance they don't know how to protect themselves from malware.
Even though I'm primarily on Linux, I collaborate a lot, and I don't want to be the one to infect anyone else.
It doesn't prevent forking per se, but it encourages collaboration, and over time, a quasi software-darwinism. In other words, the best "fork" (most stable, useful, featureful, etc) will inevitably come out on top: we all win.
If some proprietary company comes along and swoops up a bunch of BSD code and keeps their resulting product's code closed, our open source community doesn't benefit at all. We're left to re-invent the wheel, even though it was our effort that enabled their product's very existence.
It's already been said here in this commentary: BW, MS, and any other proprietary company are just looking for ways in which they can benefit from the momentum of open-source development without having to participate in the process.
I, for one, find this attitude clearly demonstrates a fairly sketchy sense of ethics. But I can't say I'm in the least bit surprised.
First, let me compliment your thorough posting. Your concise definitions are easy to understand. However, let me add to your commentary with my own...
Migration is how much it will cost to move your systems over to a new type of system. This includes any new required hardware, consultants to do the more tricky application migrations, the businesss impact of migration-related downtime and reduced productivity meshing with the learning curve, training costs to train your users and your admins to run the new system, among other things.
What you didn't mention here (though you most certainly are aware of) is that migration from NT4 to XP has costs in training, hardware, and software. A Linux solution may very well be able to leverage some of the older hardware that XP would choke on. I would say personally that migration is probably neck-and-neck in this instance, if not in favor of Linux.
Ongoing costs are the long-term items, like warranty requirements, support contacts, paying your staff
(remember, a Linux admin makes more than a Windows admin -- that makes Linux more expensive in that area). Another important calculation is reliability of the whole system, defined by mean time between failures in relationship to the mean time to repair, and the business impact of downtime associated with that.
Also remember that it is widely known that Linux admins can manage a significantly higher number of machines per admin (50-100 versus 10-20). For a small operation, this makes Windows admins cheaper, but once you begin to scale upwards in numbers of machines, the Windows admin salary numbers go through the roof.
The big trick is that TCO calculations are very subjective in nature. The criteria change from consulting firm to consulting firm, and the values used to support some of those numbers can come from a variety of different sources, and may be of questionable value.
This is precisely why Microsoft funding this study is of both questionable value and questionable ethics.
What you have to remember about IT is that it is not about the technology -- it is about the people and business that technology supports.
Ya know, I really wish more IT people understood this. Our nasty reputation gives a lot of the users we support the preconception that we're only in it for the machines.
These are the folks you keep on Windows + Office. The majority of users, however, are probably just fine on Open Office (either on Windows or Linux). It's very typical of Microsoft to make sweeping generalizations of what an organization needs (everyone on Windows XP Pro with Office 2003).
While it's unreasonable to have a personalized configuration for every last user, having half a dozen to a dozen configurations is well within even a small IT department's capability, and will likely make just about everyone happy while saving money.
And furthermore, how many of them get sent to training for each incremental MS Office version? There tends to be about as many differences between any given version of MS Office as there are between Open Office.
People!!! use "sudo". It's really not that hard to not use root. You're tempting fate if your systems allow root logins to anything but the console. At the very least, use "su".
After having received several dozen infected email messages in my inbox this week, I cannot believe that anyone would casually suggest that any yahoo should be able to set up a mail server (sorry, I don't mean your mom or grandmom is a yahoo).
The damage is NOT minimal; look at the reports of worms, trojans, etc., all spread by poorly configured mail servers.
One of Microsoft's biggest security weaknesses (in my opinion) is the fact that they market their wares as being easy enough for a yahoo to set up and run.
I don't know how anyone could feel that their network infrastructure is safe if their IT staff gets nervous at the mention of text files and the command line, and god-forbid, actually knowing what they're doing.
I guess I better install Windows on all these PCs I have here at HOME. Especially the one I'm sitting at (that has a desktop environment and a bunch of desktop apps). I haven't used Windows on the deskop of my _DESKTOP_ PCs in years. Linux can be whatever the user wants it to be, Troll.
I love how Gosling considers a license that prevents proprietary software developers from brazenly ripping off your code (GPL) to be viral. Sun's mingling with MS isn't very surprising, but it is disappointing.
I think RMS' views tend to be a bit extreme, but Gosling shows himself to be firmly planted on the opposite side of the spectrum as just another cog in the corporate greed wheel.
With this latest MS arrangement, Sun has effectively made any desire I had to learn Java pretty much go away. I've been hearing great things about Python...
I don't believe for a moment that all factors have been taken into account in this study. For example, they make the main competitor for Win+SQL Server to be Linux+Oracle -- most shops I know of use either MySQL or PostgreSQL -- a better study would have compared Win32+Oracle and Linux+Oracle. But even if the study was perfectly valid and honest, the one aspect of Linux that's noticeably not mentioned (and that is probably the key factor in anyone's decision to deploy Linux or even BSD) is the openness and freedom.
Sure there's the usually GPL bullshit FUD that's always mentioned, but my answer to that has always been: if you rely on the code that someone else has written as the core for your own work, either A) pony up and publish your results, or B) write your own damn code.
I generally think of RMS as somewhat of an extremist (many good ideas, but some pretty whacked-out ones mixed in there too), but to me the GPL is pretty simple 'cause it always comes back to your own choices and decisions. Don't like it? Don't use it.
I don't believe that running Linux is more expensive, but even it was, I think the people that have chosen that route wouldn't change, simply because the non-financial related freedoms that Linux/BSD bring are conspicuously missing from Windows. It really seems that MS hates that people actually now have the freedom of choice.
How is it that a significant chunk of code (i.e. enough code to make it very unlikely that two independent programmers happened upon the same solution, line-by-line), would even work in the kernels of two different operating systems? I realize Linux is UNIX-like, but I fail to see how this would work without at least some re-working of the supposed-stolen code.
SCO sucks dirty mouse balls! -- and for some reason, this whole thing gives me a renewed hatred for Novell/Netware...(yeah, I know it's completely off-topic, but I just felt the need to share that...)
I do go for the bleeding edge in some regard, but who doesn't want the newest bells and whistles when they are made out to sound like they'll actually promote productivity in the studio?
Here's the problem with going back to the PC: there is at least one feature in Logic 6 that I have been craving since 1998 -- non real-time rendering of a final mix. Every other app I know of has done this for years, and I've been hoping it would show up in Logic. I have the Mac now, it's mostly working, I just don't like the user experience at all. I may or may not get over it -- at this point, I guess Logic is worth working with an irritating OS to me. The cost (money and time) of investing in and learning a new package will not likely reduce any frustration at this point.
I guess I was hoping that the eMagic team had secretly developed a Linux port and were waiting for the perfect opportunity to announce it on Slashdot... (...along with Apple's port of OS X to x86 hardware)
I think what it comes down to is that I'm frustrated with the Mac. I never wanted to migrate, but because Logic was bought-out by Apple, I had the choice to either switch to Apple or find another program. At the time, I had the money for a Mac, I really love Logic Audio, and all the wonderful advertising about the great Mac experience got the better of me.
At this point, I feel like a victim of marketting hype, because when I bought the Mac, once a week, it seemed as though something would just decide to stop working right. Granted, after much troubleshooting, I've finally gotten it to a point where it works without complaining. I'm still nowhere near being a convert.
This will probably piss someone off, but my impression of the Mac is like Jack Nicholson's take on women from As Good As It Gets:
Macs are like PCs, just take away speed and organized file management...
Seriously, what I find is that even though I sequence and record on the Mac, I still heavily rely on the PC for audio editing and processing. I own a license to Sound Forge and see no compelling reason not to use it. I honestly don't even know if there is a comparable product on the Mac. But if it comes down to having to buy more software, I think that I would be more likely to buy a PC recording and sequencing package than a high-end sample editor for the Mac. The PC has more horsepower for the buck, and when it comes to real-time effects and sound-editing, it really makes a difference to have that speed boost.
BTW, I feel that it should be mentioned that I'm using OS X. Because I'm still using a PC for audio editing, I need the networking that OS X can provide. I had fewer problems with OS 9, but using it was like being forced to stare at an ugly girl after having seen a beautiful one (...and also the networking incompatibilities...).
HTTPS uses more network traffic versus HTTP, and it uses more processor time on your web server(s), at least for those of us without the luxury of hardware crypto-accelerators. I'm all for heightened security, but the last thing any of us needs is a slower, more congested Web.
While I agree that the base box alone wouldn't go for $4500, there is a $4599 model on Apple's page -- "The Ultimate". While it's basically just a tricked out version of the $3299 model (more RAM, better video card), it's listed there, and it doesn't even include any of their $1000 17" LCD monitors. Whether they're slow or not, you can't really debate that they're damned expensive for the performance you get.
I bought a Mac out of quasi-necessity; I am a musician and I have been using Emagic Logic Audio for a number of years. Apple recently bought Emagic, and naturally, PC support went out the window (no pun intended).
I also have two other systems at home; a dual PIII 800 and a dual Athlon 1.2 GHz. I wrote a quick PHP script to measure the execution time on a loop that calculates primes between two fixed numbers, and I ran it on the Mac, my Linux server, and my Linux workstation.
Here's the results (average of three runs):
dual PIII 800 MHz w/ RH Linux 7.3: 9.3 sec
dual Athlon 1.2 GHz w/ RH Linux 8.0: 6.4 sec
dual G4 1.25 GHz w/ MacOS 10.2.1: 10.6 sec
Perhaps it's just that PHP isn't as fast on an OS X box, but I basically used untuned, default installs on all three machines. The numbers were the same on the Linux boxen with or without X running.
For me, the Mac feels slower, and to me, my quick and dirty benchmark only confirms what I feel.
From what I've understood from the other posts on this topic, the biggest problem with the scenario you mention seems to be when a primary machine hangs but mysteriously recovers.
When the primary hangs, the secondary takes over (figuring that the primary is down) and mounts the drive. But when the primary recovers from the hang, it still has the drive mounted, so both systems have the drive mounted.
I believe they call that a "Bad Thing" (or massive FS corruption -- your choice).
There seems to be a bunch of ways around this scenario -- GFS/openGFS, STONITH switches, etc. -- but this is why it's not such a good idea to just let the secondary take over without being totally sure that the primary is DEAD.
OK, so as soon as I finished posting, I did a search for firewire enclosures. Wouldn't it figure? Every single freaking box had two ports. That'll teach me to post without researching...
So if it's mostly a software issue, would something like a firewire RAID box and GFS/OpenGFS work?
Slashdot Mirror
I definitely agree that knowing both is only a benefit to a good sysadmin. However, I think as time goes on, you tend towards one or the other.
The problem with wearing both caps is that you end up being a jack of trades serving two masters. (How's that for screwed-up, mixed metaphor?)
Link to a TCO study by IBM
That actually IS part of the stupid "get the facts straight" campaign. What this and other silly comparisons always conveniently leave out is the fact that a good Linux admin can handle 50-100 boxes (more in an identical cluster situation), whereas a good windows admin can handle 20-30 boxes.
How about some math:
(these salaries are made up and meant to illustrate that even if your Linux admin makes DOUBLE what your Windows admins make, it's still a better deal.)
Linux admin: $80K, Win admin: $40K
30 machines: Linux $80K, Win $40K
30-100: Linux $80K, Win $160K
100-200: Linux $160K, Win: $280K
So now let's try it with closer to real figures:
Linux admin: $50K, Win admin: $45K
30 boxes: Linux $50K, Win $45K
30-100: Linux $50K, Win $180K
100-200: Linux $100K, Win $315
If Windows admins are in fact paid less, Windows can only come out ahead in small operations. Furthermore, the closer Win and Linux admin salaries are, the better the savings are in a large operation when you go with Linux.
Also as an admin, if you see that even Microsoft is saying that Linux admins make more money, why would you waste your time learning their system? Show me the MONEY!
I request that any word documents sent to me via email are reformatted as plain text, RTF, or PDF, with my preference being plain text.
I delete any incoming Word docs as a precautionary measure; I don't feel like there's any supporting evidence why I _should_ accept Word docs. If the user can't figure out how to reformat the document, I feel that there's a damn good chance they don't know how to protect themselves from malware.
Even though I'm primarily on Linux, I collaborate a lot, and I don't want to be the one to infect anyone else.
It doesn't prevent forking per se, but it encourages collaboration, and over time, a quasi software-darwinism. In other words, the best "fork" (most stable, useful, featureful, etc) will inevitably come out on top: we all win.
If some proprietary company comes along and swoops up a bunch of BSD code and keeps their resulting product's code closed, our open source community doesn't benefit at all. We're left to re-invent the wheel, even though it was our effort that enabled their product's very existence.
It's already been said here in this commentary: BW, MS, and any other proprietary company are just looking for ways in which they can benefit from the momentum of open-source development without having to participate in the process.
I, for one, find this attitude clearly demonstrates a fairly sketchy sense of ethics. But I can't say I'm in the least bit surprised.
Sorry, I always get James Gosling and Bill Joy confused. My bad. Sorry Dr. Joy!
I think the GPL is very cool too. It's a shame that it was born out of Bill Joy screwing RMS over.
First, let me compliment your thorough posting. Your concise definitions are easy to understand. However, let me add to your commentary with my own...
What you didn't mention here (though you most certainly are aware of) is that migration from NT4 to XP has costs in training, hardware, and software. A Linux solution may very well be able to leverage some of the older hardware that XP would choke on. I would say personally that migration is probably neck-and-neck in this instance, if not in favor of Linux.
Also remember that it is widely known that Linux admins can manage a significantly higher number of machines per admin (50-100 versus 10-20). For a small operation, this makes Windows admins cheaper, but once you begin to scale upwards in numbers of machines, the Windows admin salary numbers go through the roof.
This is precisely why Microsoft funding this study is of both questionable value and questionable ethics.
Ya know, I really wish more IT people understood this. Our nasty reputation gives a lot of the users we support the preconception that we're only in it for the machines.
Peace
These are the folks you keep on Windows + Office. The majority of users, however, are probably just fine on Open Office (either on Windows or Linux). It's very typical of Microsoft to make sweeping generalizations of what an organization needs (everyone on Windows XP Pro with Office 2003).
While it's unreasonable to have a personalized configuration for every last user, having half a dozen to a dozen configurations is well within even a small IT department's capability, and will likely make just about everyone happy while saving money.
And furthermore, how many of them get sent to training for each incremental MS Office version? There tends to be about as many differences between any given version of MS Office as there are between Open Office.
People!!! use "sudo". It's really not that hard to not use root. You're tempting fate if your systems allow root logins to anything but the console. At the very least, use "su".
[end rant]
After having received several dozen infected email messages in my inbox this week, I cannot believe that anyone would casually suggest that any yahoo should be able to set up a mail server (sorry, I don't mean your mom or grandmom is a yahoo).
The damage is NOT minimal; look at the reports of worms, trojans, etc., all spread by poorly configured mail servers.
One of Microsoft's biggest security weaknesses (in my opinion) is the fact that they market their wares as being easy enough for a yahoo to set up and run.
I don't know how anyone could feel that their network infrastructure is safe if their IT staff gets nervous at the mention of text files and the command line, and god-forbid, actually knowing what they're doing.
ConceptDraw on OSX is pretty cool too. I asked them if they were pondering a Linux port, and they said they were considering it.
I guess I better install Windows on all these PCs I have here at HOME. Especially the one I'm sitting at (that has a desktop environment and a bunch of desktop apps). I haven't used Windows on the deskop of my _DESKTOP_ PCs in years. Linux can be whatever the user wants it to be, Troll.
I love how Gosling considers a license that prevents proprietary software developers from brazenly ripping off your code (GPL) to be viral. Sun's mingling with MS isn't very surprising, but it is disappointing.
I think RMS' views tend to be a bit extreme, but Gosling shows himself to be firmly planted on the opposite side of the spectrum as just another cog in the corporate greed wheel.
With this latest MS arrangement, Sun has effectively made any desire I had to learn Java pretty much go away. I've been hearing great things about Python...
I don't believe for a moment that all factors have been taken into account in this study. For example, they make the main competitor for Win+SQL Server to be Linux+Oracle -- most shops I know of use either MySQL or PostgreSQL -- a better study would have compared Win32+Oracle and Linux+Oracle. But even if the study was perfectly valid and honest, the one aspect of Linux that's noticeably not mentioned (and that is probably the key factor in anyone's decision to deploy Linux or even BSD) is the openness and freedom.
Sure there's the usually GPL bullshit FUD that's always mentioned, but my answer to that has always been: if you rely on the code that someone else has written as the core for your own work, either A) pony up and publish your results, or B) write your own damn code.
I generally think of RMS as somewhat of an extremist (many good ideas, but some pretty whacked-out ones mixed in there too), but to me the GPL is pretty simple 'cause it always comes back to your own choices and decisions. Don't like it? Don't use it.
I don't believe that running Linux is more expensive, but even it was, I think the people that have chosen that route wouldn't change, simply because the non-financial related freedoms that Linux/BSD bring are conspicuously missing from Windows. It really seems that MS hates that people actually now have the freedom of choice.
...just my opinions...
h = Planck constant (The fundamental constant equal to the ratio of the energy of a quantum of energy to its frequency. It is the quantum of action.)
E = the energy of a photon
nu = frequency of E
Planck equation: E = h*nu
therefore: nu = E/h
How is it that a significant chunk of code (i.e. enough code to make it very unlikely that two independent programmers happened upon the same solution, line-by-line), would even work in the kernels of two different operating systems? I realize Linux is UNIX-like, but I fail to see how this would work without at least some re-working of the supposed-stolen code.
SCO sucks dirty mouse balls! -- and for some reason, this whole thing gives me a renewed hatred for Novell/Netware...(yeah, I know it's completely off-topic, but I just felt the need to share that...)
I do go for the bleeding edge in some regard, but who doesn't want the newest bells and whistles when they are made out to sound like they'll actually promote productivity in the studio?
Here's the problem with going back to the PC: there is at least one feature in Logic 6 that I have been craving since 1998 -- non real-time rendering of a final mix. Every other app I know of has done this for years, and I've been hoping it would show up in Logic. I have the Mac now, it's mostly working, I just don't like the user experience at all. I may or may not get over it -- at this point, I guess Logic is worth working with an irritating OS to me. The cost (money and time) of investing in and learning a new package will not likely reduce any frustration at this point.
I guess I was hoping that the eMagic team had secretly developed a Linux port and were waiting for the perfect opportunity to announce it on Slashdot... (...along with Apple's port of OS X to x86 hardware)
I can dream, can't I?...
I think what it comes down to is that I'm frustrated with the Mac. I never wanted to migrate, but because Logic was bought-out by Apple, I had the choice to either switch to Apple or find another program. At the time, I had the money for a Mac, I really love Logic Audio, and all the wonderful advertising about the great Mac experience got the better of me.
At this point, I feel like a victim of marketting hype, because when I bought the Mac, once a week, it seemed as though something would just decide to stop working right. Granted, after much troubleshooting, I've finally gotten it to a point where it works without complaining. I'm still nowhere near being a convert.
This will probably piss someone off, but my impression of the Mac is like Jack Nicholson's take on women from As Good As It Gets:
Seriously, what I find is that even though I sequence and record on the Mac, I still heavily rely on the PC for audio editing and processing. I own a license to Sound Forge and see no compelling reason not to use it. I honestly don't even know if there is a comparable product on the Mac. But if it comes down to having to buy more software, I think that I would be more likely to buy a PC recording and sequencing package than a high-end sample editor for the Mac. The PC has more horsepower for the buck, and when it comes to real-time effects and sound-editing, it really makes a difference to have that speed boost.
BTW, I feel that it should be mentioned that I'm using OS X. Because I'm still using a PC for audio editing, I need the networking that OS X can provide. I had fewer problems with OS 9, but using it was like being forced to stare at an ugly girl after having seen a beautiful one (...and also the networking incompatibilities...).
In any case, thanks for the reply.
HTTPS uses more network traffic versus HTTP, and it uses more processor time on your web server(s), at least for those of us without the luxury of hardware crypto-accelerators. I'm all for heightened security, but the last thing any of us needs is a slower, more congested Web.
While I agree that the base box alone wouldn't go for $4500, there is a $4599 model on Apple's page -- "The Ultimate". While it's basically just a tricked out version of the $3299 model (more RAM, better video card), it's listed there, and it doesn't even include any of their $1000 17" LCD monitors. Whether they're slow or not, you can't really debate that they're damned expensive for the performance you get.
I bought a Mac out of quasi-necessity; I am a musician and I have been using Emagic Logic Audio for a number of years. Apple recently bought Emagic, and naturally, PC support went out the window (no pun intended).
I also have two other systems at home; a dual PIII 800 and a dual Athlon 1.2 GHz. I wrote a quick PHP script to measure the execution time on a loop that calculates primes between two fixed numbers, and I ran it on the Mac, my Linux server, and my Linux workstation.
Here's the results (average of three runs):
Perhaps it's just that PHP isn't as fast on an OS X box, but I basically used untuned, default installs on all three machines. The numbers were the same on the Linux boxen with or without X running.
For me, the Mac feels slower, and to me, my quick and dirty benchmark only confirms what I feel.
FWIW, Here's the code:
<?php
function timenow() {
list($microsec, $sec) = explode(" ",microtime());
return ($microsec + $sec);
}
$count = 0; //arbitrary starting point //arbitrary ending point
echo "Calculating...\n\n";
$loopstart = 6000;
$loopend = 7000;
$start = timenow();
if ($notprime == false) $count++;for ($x = $loopstart; $x <= $loopend; $x++) {
$notprime = false;
}
$end = timenow(); ." seconds to calculate\n";
echo "There are $count primes between $loopstart and $loopend\n";
echo "This took ". ($end - $start)
?>
From what I've understood from the other posts on this topic, the biggest problem with the scenario you mention seems to be when a primary machine hangs but mysteriously recovers.
When the primary hangs, the secondary takes over (figuring that the primary is down) and mounts the drive. But when the primary recovers from the hang, it still has the drive mounted, so both systems have the drive mounted.
I believe they call that a "Bad Thing" (or massive FS corruption -- your choice).
There seems to be a bunch of ways around this scenario -- GFS/openGFS, STONITH switches, etc. -- but this is why it's not such a good idea to just let the secondary take over without being totally sure that the primary is DEAD.
OK, so as soon as I finished posting, I did a search for firewire enclosures. Wouldn't it figure? Every single freaking box had two ports. That'll teach me to post without researching...
So if it's mostly a software issue, would something like a firewire RAID box and GFS/OpenGFS work?