Can you imagine being a competitor of MS's and running IIS? Microsoft could just walk right into your web server and steal damned near anything they want. Gods help you if they found a way around your firewall. But of course, Microsoft has been so ethical in the past that I'm sure they'd never do something like that...
So what you're saying is you're upset that this fancy car with the automatic transmission doesn't have a clutch? what? Look, just nfs mount an rpm directory for those boxes, then go to each one and do a: rpm -Uvh/mnt/nfs/rpms/* and you're DONE. Each machine will have its own sane RPM database, and you can check each one individually to make sure they installed correctly. After that, you can cron a script to run at night that'll do it for ya and email you the results... Let the machine do the work, not you.
1: We don't need GUI config tools, we need quality text based config tools like linuxconf, which has matured nicely in the last year or so. I do NOT want to have to wander all over the company campus to configure my servers because I can't do it via telnet. I want a simple, easy to use text based config tool for servers. I have that. It's called linuxconf. 2: The IP stack is currently "multi-threaded" (although that's the wrong terminology, it's functionally equivalent." The biggest issue right now is with PCI NICs of different chipsets not getting along. 3: Suse is shipping ReiserFS with their latest Distro now. ResierFS is REAL journaling (i.e. data and meta-data) while NT's journaling is meta-data only. I.e. turn off the power during a write, and poof, the data is gone, but the disk structure is OK. 4: As for performance management tools, I just cat or tail -f the appropriate sections of/proc. You don't need pretty graphs and clicking buttons to monitor a server, but they are nice. 5: Scalability is fine on the 2.2. kernels for file and print serving, as well as intranet web serving, where I/O bottle necks, not CPU bottlenecks are usually the problem. On database performance, such as SAP R3, Linux is pretty much equal to NT on a four way machine. Real Scalability comes in clusters, and that's an area Linux could grow a little more in, but by using DFS (distributed file system, thanks Microsoft) with several servers, you can load balance by hand a dozen linux servers that look to joe average like one. The fact is, the MOST important criteria for network OSes is reliabilty. If it isn't reliable, it doesn't even deserve to be in the competition, and Windows ANYTHING, has not proven itself worthy of the title "stable and reliable."
I'm a bisexual, polyamorous pagan. There are many people for whom I am on the short list of "people who the government could haul away and I wouldn't say anything."
I think the hate crime legislation is wrong mainly because it tries to take the decision of sentence length away from the judge and jury and tries to give it a pre-defined standard that cannot consider all the possibilities.
I prefer the jury and judge have large leeway in sentencing of crimes based on ANY motive or circumstance, not just one or two.
I imagine that most of us would agree that someone who killed hundreds of school kids just for fun should be put away forever, or to death. Another man kills a man he finds in bed with his wife in rage. Whould that guy be just as screwed if the other man happened to be another race? I could see racially motivated sentencing guidelines becoming 1984ish in nature over time.
My guess is at the time of the test their quad Xeon machine ran about $50,000, now down to $30,000 or so.
For $1000 to $2000 per box, I can build a Linux eddieware system that has redundancy, reliability, performance, and REAL scalability. Let's all just sit back and imagine a web server farm with 20 times the horsepower of slashdot taking on that poor little quad Xeon machine.
Nah, NT is the A5 vigilante, a large, underpowered attack aircraft the navy had in the 1960s. It was one of those planes that you weren't sure when you launched one if it was gonna make it back.
On the other hand, it had tons of technologically advanced electronics and lots of hype from the manufacturer.
Linux would HAVE to be the stealth fighter.
Let's see, the stealth fighter was designed on 100 year old mathmatical theories, from spare parts from other air craft (landing gear A7, engines F18, nearly every LRU in it came from some other airplane) because, hey what better way is there to hide your new server / airplane from the budget folks if you build most of it from old stuff?
Correction on the Star Fighter. It had excellent turning capabilities, was called the F-104, and it's main failing was that, unlike the F4, it NEVER shipped with a cannon, carrying only two short range missles. Maybe you meant F0192 Delta Dagger or F106 Delta Dart? They were fast little things, that turned much like a Mack Truck. But they did carry air to air nuclear tipped Genie missles. Now those things could get your attention!
Performance on CGI is far more dependent on whether or not the web server is forking for each CGI request.
I.e. if you write a.cgi program in C++, and force the server to fork for every instance that someone hits the program, you will be much slower than if you write a module in just about any language.
I would strongly suggest writing some simple benchmarks to compare what you want to do in C as a CGI, C as a module, Perl as a CGI, Perl as a module, embPERL, FastCGI etc... (and PHP4/Zend.)
You will likely find that PHP4 keeps up with almost all of them, and provides a strangely C like (and perl like as well. hmm) environment to code in.
Plus PHP now includes session management, something I'd hate to have to write in C by myself not that PHP seems so good at it.
Perl, however, is still the duct tape of the internet for the reason that many programmers know it, and it produces robust code easily and quickly.
If you use FastCGI, the perl scripts are loaded and interepreted once, then executed as binary from then on, removing the performance advantage C would have as a module (mostly) and preventing forking.
If you're more comfortable with C, or writing a truly large application, I'd strongly recommend writing your own custome modules in it.
I'm a poly kinky pagan bi geek boy, and I have several girlfriends:
Two are computer programmers. One's a marketer. One's an artist and a mother. Two are of the "beauty" type this guy says aren't interested in me. One has enhanced breasts. None of them are old fashioned girls (i.e. door mats.) They all have their own means of income, and on the average, they out earn me, although I do barely out earn a few. I don't have to spend my week earning enough money to "get lucky" with them. With or without their makeup, NONE of them are plain. (Note that they are in more than one catagory above.)
I cook dinner for the at least as much as they do for me. I cook dinner for them and their other boyfriends too!
Most important of all, I've opened my eyes to realize there are plenty of beautiful women, a fair number of them geek girls, a fair number not. Some of then find me attractive some don't and vice versa.
But if you can find the polyamorous ones, then you've got something!.
You should really read "The Microsoft File" by Wendy Goldman. You would find out that:
They invite companies to their campus with the lure that they want to buy them and their technology. They get the company to sign "sharing" agreements where they show MS all their secrets and source code. MS then says, sorry, we aren't interested, steals the ideas and builds their own product. Most of these companies are too small to mount a legal challened in court and disappear within a year or two of their MS meeting.
They openly forced most large computer sellers like Compaq and Dell to pay for windows for each and every computer they sold, even if the customer wanted a "blank" machine with no OS.
They forced most large computer sellers to bundle MS Office with Windows 95, actually charging more for just widnows 95 than for windows 95 and MS office. Ever wonder why Office is the dominant suite in the market? MS gave it away long enough to establish a lead, then hopped up the price once the competition was mostly dead.
When people were first signing up for early development kits, they had to sign forms stating they would not develop applications for other "drag and drop" interfaces with the Microsoft tools. At the time there were several alternative OLE like options, most of which were far nicer than the on in WIN32 (what a cranky, buggy interface to program).
Microsoft got where it is today by strong arming the cometition and abusing its monopoly power, whether you can see it or not.
If a Japanese firm acted like Microsoft, we'd have stopped them long ago, but because they are an American company, they have gotten away with this too long. Please stop them now before it's too late.
In any case, the scalability of Intel-based SMP systems is currently limited by the memory bus, which is so narrow (800MB/s peak) that a single CPU can saturate it.
Doesn't the AMD K7 use the Alpha switched bus architechture for SMP? I'd like to see what Linux on a four way K7-1000 will do in a years time!
Is there any advantage in Linux to RAID 1+0 over a three or four drive mirror set? (Linux supports more than 2 drives in a mirror, and reads round robin from them.) So, would 3 or 4 drives in a single mirror set generally be faster but have less storage than the 1+0 setup? Especially for a database with a majority (>90%) of reads.
In 1989 I'd been running a 32 bit os that multitasked and had multimedia capabilities for over 2 years. and could do it in 512k or 1 Meg and had good performance on a 7.2MHz CPU.
But back then it was a game machine. Sheesh. The Amiga had everything but Unix beat. No one noticed. Shame it had such a shitty display.
But the only way to achieve portability is to develop for more than one platform at a time.
Funny. Linux was NOT designed with portability in mind, using a monolithic kernel. Yet by using a fairly bland set of assumptions about uProcessor design, it runs on nearly everything (didn't I just see something about it being on the dreamcast machines now???)
Meanwhile, NT runs on TWO platforms. and 64 bit NT will be a complete rewrite! Dear God man, if you can't even port to a 64 bit CPU, how transportable can your OS be?
Slashdot Mirror
Can you imagine being a competitor of MS's and running IIS? Microsoft could just walk right into your web server and steal damned near anything they want. Gods help you if they found a way around your firewall. But of course, Microsoft has been so ethical in the past that I'm sure they'd never do something like that...
I can see it now, grandma compiling apache with -O9.
I don't know. At 1600x1200 on my 21" monitor, with the fonts cranked up to 24 or 28, they look pretty clean. ;^)
So what you're saying is you're upset that this fancy car with the automatic transmission doesn't have a clutch? what? Look, just nfs mount an rpm directory for those boxes, then go to each one and do a: rpm -Uvh /mnt/nfs/rpms/* and you're DONE. Each machine will have its own sane RPM database, and you can check each one individually to make sure they installed correctly. After that, you can cron a script to run at night that'll do it for ya and email you the results... Let the machine do the work, not you.
1: We don't need GUI config tools, we need quality text based config tools like linuxconf, which has matured nicely in the last year or so. I do NOT want to have to wander all over the company campus to configure my servers because I can't do it via telnet. I want a simple, easy to use text based config tool for servers. I have that. It's called linuxconf. 2: The IP stack is currently "multi-threaded" (although that's the wrong terminology, it's functionally equivalent." The biggest issue right now is with PCI NICs of different chipsets not getting along. 3: Suse is shipping ReiserFS with their latest Distro now. ResierFS is REAL journaling (i.e. data and meta-data) while NT's journaling is meta-data only. I.e. turn off the power during a write, and poof, the data is gone, but the disk structure is OK. 4: As for performance management tools, I just cat or tail -f the appropriate sections of /proc. You don't need pretty graphs and clicking buttons to monitor a server, but they are nice. 5: Scalability is fine on the 2.2. kernels for file and print serving, as well as intranet web serving, where I/O bottle necks, not CPU bottlenecks are usually the problem. On database performance, such as SAP R3, Linux is pretty much equal to NT on a four way machine. Real Scalability comes in clusters, and that's an area Linux could grow a little more in, but by using DFS (distributed file system, thanks Microsoft) with several servers, you can load balance by hand a dozen linux servers that look to joe average like one. The fact is, the MOST important criteria for network OSes is reliabilty. If it isn't reliable, it doesn't even deserve to be in the competition, and Windows ANYTHING, has not proven itself worthy of the title "stable and reliable."
I think the hate crime legislation is wrong mainly because it tries to take the decision of sentence length away from the judge and jury and tries to give it a pre-defined standard that cannot consider all the possibilities.
I prefer the jury and judge have large leeway in sentencing of crimes based on ANY motive or circumstance, not just one or two.
I imagine that most of us would agree that someone who killed hundreds of school kids just for fun should be put away forever, or to death. Another man kills a man he finds in bed with his wife in rage. Whould that guy be just as screwed if the other man happened to be another race? I could see racially motivated sentencing guidelines becoming 1984ish in nature over time.
Let the jury decide, not congress.
My guess is at the time of the test their quad Xeon machine ran about $50,000, now down to $30,000 or so.
For $1000 to $2000 per box, I can build a Linux eddieware system that has redundancy, reliability, performance, and REAL scalability. Let's all just sit back and imagine a web server farm with 20 times the horsepower of slashdot taking on that poor little quad Xeon machine.
There, doesn't that feel better?
Nah, NT is the A5 vigilante, a large, underpowered attack aircraft the navy had in the 1960s. It was one of those planes that you weren't sure when you launched one if it was gonna make it back.
On the other hand, it had tons of technologically advanced electronics and lots of hype from the manufacturer.
Linux would HAVE to be the stealth fighter.
Let's see, the stealth fighter was designed on 100 year old mathmatical theories, from spare parts from other air craft (landing gear A7, engines F18, nearly every LRU in it came from some other airplane) because, hey what better way is there to hide your new server / airplane from the budget folks if you build most of it from old stuff?
Correction on the Star Fighter. It had excellent turning capabilities, was called the F-104, and it's main failing was that, unlike the F4, it NEVER shipped with a cannon, carrying only two short range missles. Maybe you meant F0192 Delta Dagger or F106 Delta Dart? They were fast little things, that turned much like a Mack Truck. But they did carry air to air nuclear tipped Genie missles. Now those things could get your attention!
Performance on CGI is far more dependent on whether or not the web server is forking for each CGI request.
.cgi program in C++, and force the server to fork for every instance that someone hits the program, you will be much slower than if you write a module in just about any language.
I.e. if you write a
I would strongly suggest writing some simple benchmarks to compare what you want to do in C as a CGI, C as a module, Perl as a CGI, Perl as a module, embPERL, FastCGI etc... (and PHP4/Zend.)
You will likely find that PHP4 keeps up with almost all of them, and provides a strangely C like (and perl like as well. hmm) environment to code in.
Plus PHP now includes session management, something I'd hate to have to write in C by myself not that PHP seems so good at it.
Perl, however, is still the duct tape of the internet for the reason that many programmers know it, and it produces robust code easily and quickly.
If you use FastCGI, the perl scripts are loaded and interepreted once, then executed as binary from then on, removing the performance advantage C would have as a module (mostly) and preventing forking.
If you're more comfortable with C, or writing a truly large application, I'd strongly recommend writing your own custome modules in it.
I'm a poly kinky pagan bi geek boy, and I have several girlfriends:
Two are computer programmers.
One's a marketer.
One's an artist and a mother.
Two are of the "beauty" type this guy says aren't interested in me.
One has enhanced breasts.
None of them are old fashioned girls (i.e. door mats.)
They all have their own means of income, and on the average, they out earn me, although I do barely out earn a few. I don't have to spend my week earning enough money to "get lucky" with them.
With or without their makeup, NONE of them are plain.
(Note that they are in more than one catagory above.)
I cook dinner for the at least as much as they do for me.
I cook dinner for them and their other boyfriends too!
Most important of all, I've opened my eyes to realize there are plenty of beautiful women, a fair number of them geek girls, a fair number not. Some of then find me attractive some don't and vice versa.
But if you can find the polyamorous ones, then you've got something!.
You should really read "The Microsoft File" by Wendy Goldman. You would find out that:
They invite companies to their campus with the lure that they want to buy them and their technology. They get the company to sign "sharing" agreements where they show MS all their secrets and source code. MS then says, sorry, we aren't interested, steals the ideas and builds their own product. Most of these companies are too small to mount a legal challened in court and disappear within a year or two of their MS meeting.
They openly forced most large computer sellers like Compaq and Dell to pay for windows for each and every computer they sold, even if the customer wanted a "blank" machine with no OS.
They forced most large computer sellers to bundle MS Office with Windows 95, actually charging more for just widnows 95 than for windows 95 and MS office. Ever wonder why Office is the dominant suite in the market? MS gave it away long enough to establish a lead, then hopped up the price once the competition was mostly dead.
When people were first signing up for early development kits, they had to sign forms stating they would not develop applications for other "drag and drop" interfaces with the Microsoft tools. At the time there were several alternative OLE like options, most of which were far nicer than the on in WIN32 (what a cranky, buggy interface to program).
Microsoft got where it is today by strong arming the cometition and abusing its monopoly power, whether you can see it or not.
If a Japanese firm acted like Microsoft, we'd have stopped them long ago, but because they are an American company, they have gotten away with this too long. Please stop them now before it's too late.
In any case, the scalability of Intel-based SMP systems is currently limited by the memory bus, which is so narrow (800MB/s peak) that a single CPU can saturate it.
Doesn't the AMD K7 use the Alpha switched bus architechture for SMP? I'd like to see what Linux on a four way K7-1000 will do in a years time!
Is there any advantage in Linux to RAID 1+0 over a three or four drive mirror set? (Linux supports more than 2 drives in a mirror, and reads round robin from them.) So, would 3 or 4 drives in a single mirror set generally be faster but have less storage than the 1+0 setup? Especially for a database with a majority (>90%) of reads.
Bring your big ol NT box over to my network and let me run smbtorture against it. We'll see how long it stays up.
In 1989 I'd been running a 32 bit os that multitasked and had multimedia capabilities for over 2 years. and could do it in 512k or 1 Meg and had good performance on a 7.2MHz CPU.
But back then it was a game machine. Sheesh.
The Amiga had everything but Unix beat. No one noticed. Shame it had such a shitty display.
Funny. Linux was NOT designed with portability in mind, using a monolithic kernel. Yet by using a fairly bland set of assumptions about uProcessor design, it runs on nearly everything (didn't I just see something about it being on the dreamcast machines now???)
Meanwhile, NT runs on TWO platforms. and 64 bit NT will be a complete rewrite! Dear God man, if you can't even port to a 64 bit CPU, how transportable can your OS be?
Yes, but nothing beats head from someone who's trying so hard to express their appreciation.