Please note that the concept of a "limited liability corporation" is government intervention in a major sense of the word. A thousand years of common law holds that the owners of a business are liable for all debts and deeds of that business. The limited liability corporation, a concept that dates back only to the 1870's and which is sustained only due to major government intervention, totally destroys that concept. The owners (shareholders) of limited liability corporations are NOT held liable for the deeds and debts of the business that they own.
In short, Microsoft (and all other limited liability corporations) is the result of a major government intervention into our economy. If you say that "government should keep out of business", you should also be against the government granting "artificial person" status to limited-liability corporations. After all, if government keeps out of business, they should keep out of business in ALL respects -- including areas that benefit current businesses (such as limited liability "artificial person" status for corporations).
My understanding is that he got into an argument with Linus and others when Linus refused to put a couple of his patches into the kernel, and then he went off and sulked rather than hang around and work it out.
There was a thread on the Kernel list about him a few weeks back. You might want to go to DejaNews and check it out.
Perl takes just as much memory whether it is compiled as mod_perl or loaded as a CGI. The only real difference is that you don't have to spawn a process. That speeds things up, but doesn't reduce the memory footprint.
Remember, a separate Apache gets spawned to handle each request, up to a certain limit set in your httpd.conf. Think about it, 100 copies of Apache spawned, each with the whole code bloat of Perl embedded into it...
Better have some memory handy, that's all I say:-).
They may be refusing to honor the license agreement, but they still hold copyright to the software. There is an up to $100,000 fine under federal law for each incident in which you copy software without the permission of the copyright holder. The existence (or not) of a license agreement does not change federal law regarding the rights of copyright holders.
Again: Copyright has nothing to do with licensing. Saying "This means we can copy it and sell the copies" is thus stupid. I am aghast that anybody could even suggest engaging in an illegal act like that. I thought the whole point was to point out Microsoft's breach of contract, NOT to do something just as illegal!
We investigated putting together computers based upon this board back in November. We passed, for the following reasons:
1) The price was similar to that of an Alpha.
2) For that price, you got a 32-bit machine, not a 64-bit machine.
3) For that price, you got a machine that was similar in speed to a dual Pentium II/450, for over $1500 more.
4) For that price, you got a machine that was the same price but roughly 3/4ths the speed of an Alpha 633mhz 21114, without any of the software and support that the Alpha had at that time.
In the end, counting beans kills the notion. It just isn't adequate bang for the buck.
Red Hat won't ship binary-only drivers, and they are 50% of the Linux market. SuSE has no problem with binary-only drivers (they license the non-free OSS as part of SuSE Linux), but their presence in the U.S. is still not very large (though growing). Caldera has no problem with binary drivers but Caldera sells into business markets, not into the kinds of markets that use sound cards or 3D drivers. Slackware? Debian? Forget it!
In short, I fear that Creative is going to find that having Linux drivers is no good for their sales, since nobody is going to include those drivers with their distributions and many people of the musical persuasion are uncomfortable going somewhere on the Internet to download their drivers. Is the Linux community going to get a black eye over this? You bet! When Creative discontinues Linux support because "Linux people don't use sound cards" (as verified by their FTP download logs, showing negligible downloads of their drivers), it'll make all sorts of other companies currently interested in Linux take a step backwards.
-- Eric
besides, Linux geeks need to eat too.
on
Toshiba and EULA
·
· Score: 1
Not that I'm starving, mind you, but I'm glad that not EVERYBODY builds their Linux box from scratch!
What gets me is people who claim to be Linux fans who would rather get a box from Dell or Toshiba and bitch about it rather than get a box from VA Research or LHS that they KNOW works with Linux...
-- Eric
Without good laptops? Are you joking?
on
Toshiba and EULA
·
· Score: 1
if you mount them read-only. This was with a 120gb filesystem. Obviously it's not doing those basic file system checks upon that kind of mount. I guess I'll have to see what option "check=none" does with a big filesystem next time I have one to play with (which should be shortly, with a 180gb one).
I prefer the ICP-Vortex GDT line of RAID controllers -- there's even a fibre channel model that works fine with Linux. Leonard Z is a great guy, but I like supporting vendors who support Linux -- ICP-Vortex wrote their Linux driver back in 1.3 days, supports it, and even all of their utilities run native under Linux (none of that bull about booting to DOS to configure the RAID array).
Interesting thing about ext2: mounting a 120gb partition takes about 3 minutes if you mount it read/write, but it's almost instant if you mount it read-only. Apparently it has to pre-load all that meta-data only if you intend to write to the filesystem.
e2fsck'ing that beast took over ten minutes (I don't know how much over 'cause I gave up). Formatting it in the first place took about five to eight minutes, so I aborted my e2fsck and reformatted the partition (this is while I was doing system setup and configuration, so there wasn't any data on it).
We can go up to half a terabyte without going to an external cabinet, using a solid heavy-duty steel California PC Products case and CRU LVD hot-swap backplanes rather than that effete gee-whiz stuff that's flimsy and breaks easily. This is in a dual Xeon 450 configuration. LHS also has a quad Xeon setup that has the horsepower to break the terabyte mark (dual PCI busses, etc.), it's pretty much the same thing VA Research sells (after all, there's not many providers of quad Xeon motherboards for system integrators: Intel and AMI). With commonly available 18gb drives this would require an external RAID cabinet or two. 36gb drives should be available shortly, and those will solve some of the space and heat problems (you better have a big server room for a terabyte of storage using 18gb drives!).
We have a new laptop line, but not in the $1500 price point range. We can do you a very nice one with a 15.1" LCD display and 6gb hard drive with 128mb of memory for around $2500. With Linux. And you aren't paying for Windows 98 as part of that either -- we get'em from the manufacturer without any OS.
They're kinda big and heavy, they're not fancy, and that Mobile PII sucks power, but they work quite well.
The problem with the $1500 price point is that we'd end up supporting two different lines of laptops -- a low end line, and a high-end line. And most people interested in Linux laptops would be most interested in that $2500 laptop. Due to the low margin on a $1500 laptop we'd have to sell a couple dozen of them just to make back the cost of adding it to our lineup, much less the ongoing support costs. I'm not sure it'd sell well enough to make back that cost.
Regarding traditional channels: We already advertise on Slashdot (for which we donated a 9gb Cheetah, BTW), Linux Journal, Linux Today (we were their first advertiser!), etc. The problem is that these channels are geared towards current Linux users, who are predominantly either hobbiests or server users. The hobby users don't buy machines -- they build their own from parts. The server users aren't generally in low-end IDE workstations in mass quantities. This means using different channels for moving mass quantities of low-end IDE workstations, especially if you want to address the consumer market. Linux Journal doesn't address that market. Mainstream publications do. But, as I mentioned, you have a chicken/egg problem there. I don't know if any vendor is going to go there (I don't make those kinds of business decisions for LHS and I know about as much as you do about VA Research), but so it goes.
I keep fussing at them and they keep ignoring me when I point out the things that need to be done to make SuSE Linux enterprise-ready.
Example:/etc is supposed to contain all configuration information. Well, in SuSE Linux it doesn't. For some weird reason they stuck their startup script configuration info in/sbin. For some weird reason they stuck their "X" server link in/var/lib/X11. Etc. The problem is that I can't go tracing down configuration information to be backed up all over God's creation. That's what/etc was invented for. But SuSE says that the "X" server link is an executable and thus does not belong in/etc. ARGH!
Red Hat Linux has its quality problems, but I have administered it in large-scale enterprise deployments and can attest that it scales quite well because of a number of pleasant design decisions, such as the/etc/cron.daily (etc.) directories, which allow me to 'ssh' out a cron job to be added to the crontab of the entire network with one command rather than having to edit an/etc/crontab on hundreds of machines, and/etc/profile.d, which allows me to add startup tasks to users without having to edit the global/etc/profile script on hundreds of machines. Then there is the fact that with Red Hat Linux, I know that if I back up/etc,/usr/local, and/home I can restore the entire system. With SuSE, configuration info is so scattered that you have to backup the entire system -- period.
Don't get me wrong, I've run SuSE and like it, but they have to be the most STUBBORN people I've ever encountered. I've been ragging them about this since SuSE 5.1 days, and still nada.
The problem on the client side is that margins are so slim. It takes about 2 man-hours of labor to build, test, and box a low-end IDE machine. (Yes, I've done time and motion studies on the subject, that's my job). The end result is maybe $100 profit. You have to sell a LOT of $100 profits to justify a technician's $25K/year salary or a systems specialist's $40K/year salary (note: These are North Carolina salaries, Silicon Valley salaries would be roughly double the above due to the higher rents and cost of living out there).
VA Research is addressing this by re-badging a Taiwanese import (we still build our own at LHS, but we have a lower-cost labor supply in North Carolina than VAR has in the Silicon Valley), but this still doesn't address the chicken-and-egg problem: you're not making enough money from client-side Linux to justify spending money on marketing it, but without pushing product through the channel you never WILL make money on it. Yet without making money on it you can't afford to take ad money away from pushing your servers and spend it all in PC Week or Computer Shopper pushing desktops... then you lose your server market (where both VAR and LHS can undercut Compaq handily on similar computers).
The answer, I guess, would be venture capital, and massive amounts of it, properly applied. We'll have to see whether it'll be one of the existing Linux hardware companies that does it, or a newcomer.
While I enjoy reading horror stories about VA Research bunging (see my URL for why:-), I have to point out that s**t happens. Orders can be delayed because of supply problems (for example, at the end of December you COULD NOT GET a Pentium II 350 or a 9gb Seagate Cheetah, period, unless you were Compaq or IBM). Orders can be bungled because the sales guy keeps rushing back to the build area and shouting "Get order #XYZ out, he's called AGAIN!" and a rushed tech accidentally skips a checksheet item. Or it may be that you ordered a combination that the guy who built your computer hates. For example, I *hate* dual-boot machines. Windows 98 is a pain in the a** to install, and a waste of valuable disk space (grin). So if I'm rushed, I might not QA the Win98 the way I should and a driver might not get loaded for something in Win98 (but you better believe that the Linux stuff is going to be right!). It happens, though not often.
Anyhow, the point is that s**t happens, regardless of the vendor, and the difference between a bad vendor and a good vendor is what happens once all h**l breaks loose. Or even all heck -- e.g., a very large server we shipped out with a token ring card didn't token ring very well (we'd warned this customer, a major bank with a single-digit address on Wall Street, that we had no way of testing a token ring card because we don't have a token ring network handy, all we could do was make sure it was recognized by the driver). So we worked with them until it was fixed, trying different kernel versions and trying patched drivers from the author of the token ring driver to try to figure out why it was Oopsing. The final solution was a different model token ring card, which we shipped to them overnight once I decided that was the solution.
I'm not going to comment any further on this issue, because of an obvious conflict of interest (grin).
It's sort of like cameras. You can go to Walmart to buy a camera, or you can go to a camera store. If you want just a plain vanilla camera, or you already know exactly what you want and it's at Walmart, you go to Walmart. But you buy what they have on the shelf -- they won't special-order anything for you, or let you mix and match stuff. If you want a good professional-quality outfit, you go to your local camera shop, because the people there know their cameras (unlike the droids at Walmart). There you can get exactly the camera you want, even if it's not in stock at the moment.
Compaq is the Walmart of Linux computers, while folks like VA Research and Linux Hardware Solutions are the knowledgable specialists (the camera shop of Linux computers). (Though I get a chuckle over the rest of the thread about people's troubles with VA Research:-). A LHS or VAR is never going to be as big as Walmart (or Compaq), but it is still a very profitable and useful market niche. Walmart has not driven the local camera shop out of business, and it's unlikely that Compaq is going to drive the local Linux shop out of business.
BTW, Debian, Mindspring, and LHS are about to release a press release about the new Debian FTP server. VA Research isn't the *only* company that gives to Open Source.
-- Eric
California PC Products makes good cases
on
Cooler Cases
·
· Score: 1
Heavy duty industrial stuff. We use'em for all of our high end stuff (alas, they're too expensive for our low end stuff).
They do look industrial, though -- heavy-gauge painted steel all the way. Can't change that basic industrial look no matter what color you paint it.
-- E
Case On + Lots of Fans == More Cooling
on
Cooler Cases
·
· Score: 1
Heat on the CPU doesn't bother me because I don't overclock. Heat on the hard drives, on the other hand, does bother me. A 10,000 RPM Cheetah becomes noticably hot to the touch with the cover of my case off. That's because no air is blowing over it. With my case on, it's a different story. I have three fans blowing *OUT* of my case (it's a hacked tower case), and big vents on the front that are set up to direct air through the hard drive bay. My hard drive runs *COOL*. And that's what I care about, because it is a lot more expensive than whatever crappy CPU I have in my machine today (I've always been of the opinion that anyone who spends more than $100 on a CPU would be better off getting a better hard drive and more memory!).
All of this depends on your case design, of course. For example, cooling in the InWin cases is kind of iffy. The newest InWin towers have enough space for three fans in addition to the power supply fan and that does help, but their front bezel needs some more vent space and they don't need all those open vents on the side and top. (They need some, but not as much as they have). On the other hand, this crappy no-name generic case I have here at home is a terrible case but it keeps my drives cool as ice with its big front vents blowing straight into the drive bays.
Many of the "programmers" I know who are not college educated have a very shallow view of computers. They only know Microsoft products, they only know Visual BASIC, etc. If they are uber-geek types, they will spend hours cutting cycles from within a do-loop rather than coming up with an algorithm that doesn't *NEED* a do-loop.
This isn't to say that college would have turned these sad excuses for programmers into super hackers, but at least they would have been exposed to the concept of O(n) vs O(1) algorithms (and don't forget O(n**2) algorithms!). And at least they'd have been exposed to the concepts of structured programming, rather than slinging spaghetti all over the place.
Having cleaned up after so many non-college-educated programmers during my career, I am probably biased. I'll just note that usually I end up re-writing rather than fixing programs created by non-college-educated programmers, because they are usually unreadable spaghetti with no comments, cryptic variable names, and no structure. Heck, I knew one kid who thought that indenting his code was structure! (And who wondered why I fussed at him when he re-read a table out of the database every time through the loop, rather than cacheing it before he entered the loop -- it took fewer lines of code, after all!). Some of the code I've read by college-educated programmers has been a bit bletcherous, but at least I can fix it without re-writing the whole thing.
This isn't to say that ALL non-college-educated programmers can't structure their way out of a paper bag... just that, in my experience, most of them are woefully ignorant of what really comprises programming -- creating easy-to-use, maintainable code that is easily extensible in the future. All the cycle counting in the world won't help if the program is unusable due to lousy user interface, unmaintainable due to poor structure, zero comments, and slower than mollasses because of ignorance of basic algorithms (Knuth and Sedgewick are the Patron Saints of algorithms, and anybody who dares differ on that shall be cast into the holy flames reserved for heretics!).
Now that I've ticked off every hacker without a college degree, I'll just say that most colleges don't teach what people need to know. I didn't need college to teach me HTML or Pascal or Python or "C" or 6502 assembly or etc... but I did need to know the difference between an O(n**2) algorithm and an O(log(n)) algorithm, and between a b-tree and a hash table. Yet so many colleges see their duty as "preparing kids for jobs in industry", and instead spend their time teaching young people whatever the "hot technology of the day" is -- a technology which will be long obsolete, of course, by the time you graduate.
Well, our costs on Celeron 300A's has plunged to below $70, and we now have Celeron 366's in stock for more than $150 less than Pentium II 350's... does anybody else see a problem with this picture?!
(Note, though, that the PII-350 has a 100mhz system bus, while the Celeron 366 has a 66mhz system bus, so the PII-350 still outperforms the Celeron 366 by a slight amount... but $150 worth?!).
5MB/sec sustained throughput hard?! It is child's play. I could do that with a single Seagate Cheetah 9gb hard drive without the hard drive even breathing hard. Note that ftp.cdrom.com has a 4-channel Mylex RAID controller, meaning that they can pretty much saturate the PCI bus (you can count on being able to burst at least 80mb/sec, 5mb/sec is DEFINITELY child's play!).
Also note that cdrom.com is definitely NOT using a lousy little T1 line. They have multiple DS-3's plus one of the BIG pipes (I forget what it's called, go see the original article and click on the link at the bottom that goes to ftp.cdrom.com).
Slashdot Mirror
Please note that the concept of a "limited liability corporation" is government intervention in a major sense of the word. A thousand years of common law holds that the owners of a business are liable for all debts and deeds of that business. The limited liability corporation, a concept that dates back only to the 1870's and which is sustained only due to major government intervention, totally destroys that concept. The owners (shareholders) of limited liability corporations are NOT held liable for the deeds and debts of the business that they own.
In short, Microsoft (and all other limited liability corporations) is the result of a major government intervention into our economy. If you say that "government should keep out of business", you should also be against the government granting "artificial person" status to limited-liability corporations. After all, if government keeps out of business, they should keep out of business in ALL respects -- including areas that benefit current businesses (such as limited liability "artificial person" status for corporations).
-- Eric
My understanding is that he got into an argument with Linus and others when Linus refused to put a couple of his patches into the kernel, and then he went off and sulked rather than hang around and work it out.
There was a thread on the Kernel list about him a few weeks back. You might want to go to DejaNews and check it out.
-- Eric
Perl takes just as much memory whether it is compiled as mod_perl or loaded as a CGI. The only real difference is that you don't have to spawn a process. That speeds things up, but doesn't reduce the memory footprint.
:-).
Remember, a separate Apache gets spawned to handle each request, up to a certain limit set in your httpd.conf. Think about it, 100 copies of Apache spawned, each with the whole code bloat of Perl embedded into it...
Better have some memory handy, that's all I say
-- Eric
They may be refusing to honor the license agreement, but they still hold copyright to the software. There is an up to $100,000 fine under federal law for each incident in which you copy software without the permission of the copyright holder. The existence (or not) of a license agreement does not change federal law regarding the rights of copyright holders.
Again: Copyright has nothing to do with licensing. Saying "This means we can copy it and sell the copies" is thus stupid. I am aghast that anybody could even suggest engaging in an illegal act like that. I thought the whole point was to point out Microsoft's breach of contract, NOT to do something just as illegal!
-- Eric
We investigated putting together computers based upon this board back in November. We passed, for the following reasons:
1) The price was similar to that of an Alpha.
2) For that price, you got a 32-bit machine, not a 64-bit machine.
3) For that price, you got a machine that was similar in speed to a dual Pentium II/450, for over $1500 more.
4) For that price, you got a machine that was the same price but roughly 3/4ths the speed of an Alpha 633mhz 21114, without any of the software and support that the Alpha had at that time.
In the end, counting beans kills the notion. It just isn't adequate bang for the buck.
-- Eric
Red Hat won't ship binary-only drivers, and they are 50% of the Linux market. SuSE has no problem with binary-only drivers (they license the non-free OSS as part of SuSE Linux), but their presence in the U.S. is still not very large (though growing). Caldera has no problem with binary drivers but Caldera sells into business markets, not into the kinds of markets that use sound cards or 3D drivers. Slackware? Debian? Forget it!
In short, I fear that Creative is going to find that having Linux drivers is no good for their sales, since nobody is going to include those drivers with their distributions and many people of the musical persuasion are uncomfortable going somewhere on the Internet to download their drivers. Is the Linux community going to get a black eye over this? You bet! When Creative discontinues Linux support because "Linux people don't use sound cards" (as verified by their FTP download logs, showing negligible downloads of their drivers), it'll make all sorts of other companies currently interested in Linux take a step backwards.
-- Eric
Not that I'm starving, mind you, but I'm glad that not EVERYBODY builds their Linux box from scratch!
What gets me is people who claim to be Linux fans who would rather get a box from Dell or Toshiba and bitch about it rather than get a box from VA Research or LHS that they KNOW works with Linux...
-- Eric
Are our laptops toast?
-- Eric
(Linux Hardware Solutions)
and counting :-).
-- Eric
How about RetroMan? Wears 70's duds and listens to Disco while fighting crime cruisin' in his El Camino listening to 8-track tapes?
if you mount them read-only. This was with a 120gb filesystem. Obviously it's not doing those basic file system checks upon that kind of mount. I guess I'll have to see what option "check=none" does with a big filesystem next time I have one to play with (which should be shortly, with a 180gb one).
-- Eric
since we're in blatant commercialism mode (sigh).
I prefer the ICP-Vortex GDT line of RAID controllers -- there's even a fibre channel model that works fine with Linux. Leonard Z is a great guy, but I like supporting vendors who support Linux -- ICP-Vortex wrote their Linux driver back in 1.3 days, supports it, and even all of their utilities run native under Linux (none of that bull about booting to DOS to configure the RAID array).
Interesting thing about ext2: mounting a 120gb partition takes about 3 minutes if you mount it read/write, but it's almost instant if you mount it read-only. Apparently it has to pre-load all that meta-data only if you intend to write to the filesystem.
e2fsck'ing that beast took over ten minutes (I don't know how much over 'cause I gave up). Formatting it in the first place took about five to eight minutes, so I aborted my e2fsck and reformatted the partition (this is while I was doing system setup and configuration, so there wasn't any data on it).
We can go up to half a terabyte without going to an external cabinet, using a solid heavy-duty steel California PC Products case and CRU LVD hot-swap backplanes rather than that effete gee-whiz stuff that's flimsy and breaks easily. This is in a dual Xeon 450 configuration. LHS also has a quad Xeon setup that has the horsepower to break the terabyte mark (dual PCI busses, etc.), it's pretty much the same thing VA Research sells (after all, there's not many providers of quad Xeon motherboards for system integrators: Intel and AMI). With commonly available 18gb drives this would require an external RAID cabinet or two. 36gb drives should be available shortly, and those will solve some of the space and heat problems (you better have a big server room for a terabyte of storage using 18gb drives!).
Blatant commercialism. Yetch.
-- E
We have a new laptop line, but not in the $1500 price point range. We can do you a very nice one with a 15.1" LCD display and 6gb hard drive with 128mb of memory for around $2500. With Linux. And you aren't paying for Windows 98 as part of that either -- we get'em from the manufacturer without any OS.
They're kinda big and heavy, they're not fancy, and that Mobile PII sucks power, but they work quite well.
The problem with the $1500 price point is that we'd end up supporting two different lines of laptops -- a low end line, and a high-end line. And most people interested in Linux laptops would be most interested in that $2500 laptop. Due to the low margin on a $1500 laptop we'd have to sell a couple dozen of them just to make back the cost of adding it to our lineup, much less the ongoing support costs. I'm not sure it'd sell well enough to make back that cost.
Regarding traditional channels: We already advertise on Slashdot (for which we donated a 9gb Cheetah, BTW), Linux Journal, Linux Today (we were their first advertiser!), etc. The problem is that these channels are geared towards current Linux users, who are predominantly either hobbiests or server users. The hobby users don't buy machines -- they build their own from parts. The server users aren't generally in low-end IDE workstations in mass quantities. This means using different channels for moving mass quantities of low-end IDE workstations, especially if you want to address the consumer market. Linux Journal doesn't address that market. Mainstream publications do. But, as I mentioned, you have a chicken/egg problem there. I don't know if any vendor is going to go there (I don't make those kinds of business decisions for LHS and I know about as much as you do about VA Research), but so it goes.
-- Eric
I keep fussing at them and they keep ignoring me when I point out the things that need to be done to make SuSE Linux enterprise-ready.
/etc is supposed to contain all configuration information. Well, in SuSE Linux it doesn't. For some weird reason they stuck their startup script configuration info in /sbin. For some weird reason they stuck their "X" server link in /var/lib/X11. Etc. The problem is that I can't go tracing down configuration information to be backed up all over God's creation. That's what /etc was invented for. But SuSE says that the "X" server link is an executable and thus does not belong in /etc. ARGH!
/etc/cron.daily (etc.) directories, which allow me to 'ssh' out a cron job to be added to the crontab of the entire network with one command rather than having to edit an /etc/crontab on hundreds of machines, and /etc/profile.d, which allows me to add startup tasks to users without having to edit the global /etc/profile script on hundreds of machines. Then there is the fact that with Red Hat Linux, I know that if I back up /etc, /usr/local, and /home I can restore the entire system. With SuSE, configuration info is so scattered that you have to backup the entire system -- period.
Example:
Red Hat Linux has its quality problems, but I have administered it in large-scale enterprise deployments and can attest that it scales quite well because of a number of pleasant design decisions, such as the
Don't get me wrong, I've run SuSE and like it, but they have to be the most STUBBORN people I've ever encountered. I've been ragging them about this since SuSE 5.1 days, and still nada.
-- Eric
The problem on the client side is that margins are so slim. It takes about 2 man-hours of labor to build, test, and box a low-end IDE machine. (Yes, I've done time and motion studies on the subject, that's my job). The end result is maybe $100 profit. You have to sell a LOT of $100 profits to justify a technician's $25K/year salary or a systems specialist's $40K/year salary (note: These are North Carolina salaries, Silicon Valley salaries would be roughly double the above due to the higher rents and cost of living out there).
VA Research is addressing this by re-badging a Taiwanese import (we still build our own at LHS, but we have a lower-cost labor supply in North Carolina than VAR has in the Silicon Valley), but this still doesn't address the chicken-and-egg problem: you're not making enough money from client-side Linux to justify spending money on marketing it, but without pushing product through the channel you never WILL make money on it. Yet without making money on it you can't afford to take ad money away from pushing your servers and spend it all in PC Week or Computer Shopper pushing desktops... then you lose your server market (where both VAR and LHS can undercut Compaq handily on similar computers).
The answer, I guess, would be venture capital, and massive amounts of it, properly applied. We'll have to see whether it'll be one of the existing Linux hardware companies that does it, or a newcomer.
-- Eric
While I enjoy reading horror stories about VA Research bunging (see my URL for why :-), I have to point out that s**t happens. Orders can be delayed because of supply problems (for example, at the end of December you COULD NOT GET a Pentium II 350 or a 9gb Seagate Cheetah, period, unless you were Compaq or IBM). Orders can be bungled because the sales guy keeps rushing back to the build area and shouting "Get order #XYZ out, he's called AGAIN!" and a rushed tech accidentally skips a checksheet item. Or it may be that you ordered a combination that the guy who built your computer hates. For example, I *hate* dual-boot machines. Windows 98 is a pain in the a** to install, and a waste of valuable disk space (grin). So if I'm rushed, I might not QA the Win98 the way I should and a driver might not get loaded for something in Win98 (but you better believe that the Linux stuff is going to be right!). It happens, though not often.
Anyhow, the point is that s**t happens, regardless of the vendor, and the difference between a bad vendor and a good vendor is what happens once all h**l breaks loose. Or even all heck -- e.g., a very large server we shipped out with a token ring card didn't token ring very well (we'd warned this customer, a major bank with a single-digit address on Wall Street, that we had no way of testing a token ring card because we don't have a token ring network handy, all we could do was make sure it was recognized by the driver). So we worked with them until it was fixed, trying different kernel versions and trying patched drivers from the author of the token ring driver to try to figure out why it was Oopsing. The final solution was a different model token ring card, which we shipped to them overnight once I decided that was the solution.
I'm not going to comment any further on this issue, because of an obvious conflict of interest (grin).
-- Eric
It's sort of like cameras. You can go to Walmart to buy a camera, or you can go to a camera store. If you want just a plain vanilla camera, or you already know exactly what you want and it's at Walmart, you go to Walmart. But you buy what they have on the shelf -- they won't special-order anything for you, or let you mix and match stuff. If you want a good professional-quality outfit, you go to your local camera shop, because the people there know their cameras (unlike the droids at Walmart). There you can get exactly the camera you want, even if it's not in stock at the moment.
:-). A LHS or VAR is never going to be as big as Walmart (or Compaq), but it is still a very profitable and useful market niche. Walmart has not driven the local camera shop out of business, and it's unlikely that Compaq is going to drive the local Linux shop out of business.
Compaq is the Walmart of Linux computers, while folks like VA Research and Linux Hardware Solutions are the knowledgable specialists (the camera shop of Linux computers). (Though I get a chuckle over the rest of the thread about people's troubles with VA Research
BTW, Debian, Mindspring, and LHS are about to release a press release about the new Debian FTP server. VA Research isn't the *only* company that gives to Open Source.
-- Eric
Heavy duty industrial stuff. We use'em for all of our high end stuff (alas, they're too expensive for our low end stuff).
They do look industrial, though -- heavy-gauge painted steel all the way. Can't change that basic industrial look no matter what color you paint it.
-- E
Heat on the CPU doesn't bother me because I don't overclock. Heat on the hard drives, on the other hand, does bother me. A 10,000 RPM Cheetah becomes noticably hot to the touch with the cover of my case off. That's because no air is blowing over it. With my case on, it's a different story. I have three fans blowing *OUT* of my case (it's a hacked tower case), and big vents on the front that are set up to direct air through the hard drive bay. My hard drive runs *COOL*. And that's what I care about, because it is a lot more expensive than whatever crappy CPU I have in my machine today (I've always been of the opinion that anyone who spends more than $100 on a CPU would be better off getting a better hard drive and more memory!).
All of this depends on your case design, of course. For example, cooling in the InWin cases is kind of iffy. The newest InWin towers have enough space for three fans in addition to the power supply fan and that does help, but their front bezel needs some more vent space and they don't need all those open vents on the side and top. (They need some, but not as much as they have). On the other hand, this crappy no-name generic case I have here at home is a terrible case but it keeps my drives cool as ice with its big front vents blowing straight into the drive bays.
-- Eric
It wouldn't be the first time a commercial product had been exported with one....
Remember all that key escrow stuff?
-- Eric
Many of the "programmers" I know who are not college educated have a very shallow view of computers. They only know Microsoft products, they only know Visual BASIC, etc. If they are uber-geek types, they will spend hours cutting cycles from within a do-loop rather than coming up with an algorithm that doesn't *NEED* a do-loop.
This isn't to say that college would have turned these sad excuses for programmers into super hackers, but at least they would have been exposed to the concept of O(n) vs O(1) algorithms (and don't forget O(n**2) algorithms!). And at least they'd have been exposed to the concepts of structured programming, rather than slinging spaghetti all over the place.
Having cleaned up after so many non-college-educated programmers during my career, I am probably biased. I'll just note that usually I end up re-writing rather than fixing programs created by non-college-educated programmers, because they are usually unreadable spaghetti with no comments, cryptic variable names, and no structure. Heck, I knew one kid who thought that indenting his code was structure! (And who wondered why I fussed at him when he re-read a table out of the database every time through the loop, rather than cacheing it before he entered the loop -- it took fewer lines of code, after all!). Some of the code I've read by college-educated programmers has been a bit bletcherous, but at least I can fix it without re-writing the whole thing.
This isn't to say that ALL non-college-educated programmers can't structure their way out of a paper bag... just that, in my experience, most of them are woefully ignorant of what really comprises programming -- creating easy-to-use, maintainable code that is easily extensible in the future. All the cycle counting in the world won't help if the program is unusable due to lousy user interface, unmaintainable due to poor structure, zero comments, and slower than mollasses because of ignorance of basic algorithms (Knuth and Sedgewick are the Patron Saints of algorithms, and anybody who dares differ on that shall be cast into the holy flames reserved for heretics!).
Now that I've ticked off every hacker without a college degree, I'll just say that most colleges don't teach what people need to know. I didn't need college to teach me HTML or Pascal or Python or "C" or 6502 assembly or etc... but I did need to know the difference between an O(n**2) algorithm and an O(log(n)) algorithm, and between a b-tree and a hash table. Yet so many colleges see their duty as "preparing kids for jobs in industry", and instead spend their time teaching young people whatever the "hot technology of the day" is -- a technology which will be long obsolete, of course, by the time you graduate.
-- Eric
Well, our costs on Celeron 300A's has plunged to below $70, and we now have Celeron 366's in stock for more than $150 less than Pentium II 350's... does anybody else see a problem with this picture?!
(Note, though, that the PII-350 has a 100mhz system bus, while the Celeron 366 has a 66mhz system bus, so the PII-350 still outperforms the Celeron 366 by a slight amount... but $150 worth?!).
-- Eric
Near the top, Microsoft says their site serves 6gb per day via FTP.
Near the bottom, Microsoft says that they have three quad-PPro servers dedicated solely to serving FTP. They don't do one thing more.
Pitiful, eh?
-- Eric
5MB/sec sustained throughput hard?! It is child's play. I could do that with a single Seagate Cheetah 9gb hard drive without the hard drive even breathing hard. Note that ftp.cdrom.com has a 4-channel Mylex RAID controller, meaning that they can pretty much saturate the PCI bus (you can count on being able to burst at least 80mb/sec, 5mb/sec is DEFINITELY child's play!).
Also note that cdrom.com is definitely NOT using a lousy little T1 line. They have multiple DS-3's plus one of the BIG pipes (I forget what it's called, go see the original article and click on the link at the bottom that goes to ftp.cdrom.com).
-- Eric