I admit I skimmed most of the article (looking mainly for an answer to this question), but it didn't seem to mention in the system configurations or such which file systems were being used on the Linux and Windows boxes. Since we're talking file server throughput, the file system being used would have a major effect on the "getting the stuff off the drive" part of the test. Not that this should surprise me.
if some ISP wanted to go after NAT users
If your ISP charges per computer (such as cable modem providers), then they should have a broken law on their side. You've contracted to use their service with one machine (and on a shared-bandwidth network, there's good reason for that contract limitation). If you start hooking up more machines, you're effectively splicing those machines onto the network, just like with CableTV.
If your ISP authorizes the setting up of networks to share they bandwidth they provide you, the law would not apply.
and, indeed, what he did say in his explanation is not that Moore's Law is at and end, but rather that Intel can no longer keep up with it. This is not surprising, since apparantly Intel fired all of its engineers a couple years ago and replaced them with 17 year old overclockers.
AMD, on the other hand, is finishing up a 64bit processor that suffers zero (or at worst nominal) slowdowns on systems with 32 bit OSs when compared to comparable 32 bit processors, leaving plenty of room for their power to continue to grow as the software catches up.
There are two factors (other than collusion) that could account for that phenomenon (which I've seen as well, both from the inside and the outside). First: There are certain products that make sense to advertise because of their drawing power that pretty much every store in the same category is going to use. These are the ones that have nice numbers about them, but are cheap. So, if you've got a particular 17" monitor you can sell for $79.99, you advertise that. This gets people into the store. If they want that monitor, they can, of course, buy it, but you're often hoping that they'll change their mind when they see better options and get something with a better margin. (It's only bait and switch if you don't have the product you advertise.) Second: Because consumer electronics giants are dumbshits when it comes to pricing (they'd drive themselves out of business by eliminating all margin if they could), the manufactorers place limits on what how cheaply you can advertise their products, in order to make sure their product doesn't become a money-loser and loose support of the store. (Happened to HP ink jets when I worked at CompUSA years ago. Due to Best Buy's use of them as a draw, margins were around $2.50 on them, as opposed to $30 on a Canon. Our store couldn't get HP inkjets in quantity for months.) This is called the Minimum Advertised Price, or MAP. It's semi-legal, since it's an agreement between the store and the manufactorer, not between stores, and because it doesn't prevent you from selling below the price -- only from advertising below the price. So when BB or Circuit City or whomever has that Deskjet 635 on sale for $49.95, they all end up with the same price because they're all discounting it to the MAP.
The sound being produced is not the related to the noise some hard drives make. He's not spinning them at particular rates to get different sounds (which would only allow him to produce one frequency per platter anyway). He's vibrating them, just like regular drivers. Still a cool waste of time, but not the least bit what the posting indicated.
And lets not forget...
on
Heart of the Net
·
· Score: 3, Insightful
Amid all those "outsiders of the week", ever since the net went public its biggest componants and financial backers have been the huge brick and mortar businesses. Suburban hackers and dot coms have at their best been a far second to IBM, 3M, Merrill Lynch, Wells Fargo, General Motors and all of the other multi-billion dollar companies that, through their own use of the net, covered the costs of building it so the eschewed hackers Jon so loves to think he's in touch with could write a few perl scripts.
If we're talking about ideas that are at the heart of the net and what most represents them, I'd have to say that for the last 8 or 10 years - since the net started to be really popular - that idea has been self-promotion, whether of a mega corporation or an idea or a person. My vote for the epicenter of this use of the net is jennicam.
Dead trees look better on other dead trees then CDs do (much less an ethernet port).
Being a college instructor, I need books on my shelf that people will assume I've read to improve my standing in the tribe.
Having more than two devices on a channel only causes slowdown on a broken channel design. I've had 5 drives chained onto an UW SCSI card and benchmarked better data transfer off of them while simultaneously running the benchmark on all five drives than I got from a singe 7200RPM ATA66 drive in my other system. (This was a couple years back.)
I wouldn't expect most people to realize that, of course. SCSI is, as was noted, really quite pricey. But it's damned fast and doesn't break a sweat being chained. It's unfortunate that it was never able to get into competative price points with IDE and the various kludges that have been made to it over the year. While I don't have the systems to test it, I'd be willing to bet that a fast SCSI II system with the best drives available from 1992 would still blow the doors off of a brand spanking new ATA 100 system in data transfer.
It's CmdrTaco's summary. The article only claims these are "Toys to Avoid", by whatever standards the site uses. Which begs the further question of whether CmndTaco presenting this as "News for Nerds, Stuff that Matters" is misleading as well.
Well, he's obviously old enough to remember a time when he almost had a job as a real journalist. I suspect the scanning idea. Putting in more than the absolute minimum effort needed to get crap published hasn't really been his strong suit.
Any time you see an argument purporting that Unix is easier to admin than WinNT or Novell it's coming from someone who has never spent any time administering real servers.
We've got this nifty little thing here called a SunRay. For those not familiar with them, it's a cool box from Sun that serves up Xterms to dumb terminals. We've got about a dozen terminals attached to it at the moment. While my job is not adminning it, I'm allowed to. I can do absolutely anything to absolutely anyone from any of the terminals, and when I log in, it will be configured just like my desktop, because it is my desktop. They even use these cards that will override the current session with that of the owner of the card as long as the card is inserted in the terminal, so I can go up to someone's machine, stick the card in, and immediately be working on my machine.
Naturally, I can shell into the machine from any other machine in the company, and while I won't have my pretty Xterm, I'll have the same level of control. I can do anything from move files around to fix someone's screen color and resolution.
Of course, this isn't really a good comparison, since MS doesn't even make a multi-user OS, so they don't have anything like a SunRay to take advantage of the fact that, since it's inception, their OS was built to exist as a networked OS.
Any time you see an argument purporting that WinNT/2000 is even in the same league as Unix as a server OS, much less as good or better than Unix, it's coming from someone who's never even seen, or perhaps even conceived of, real servers.
You go right ahead and consider your cute little proliant to be a bit more than a high end workstation. My $2000 game machine at home is a dual athalon with 1gb of ram running Win2K. If you want to think of something with only 4 times the processing power (assuming, almost certainly falsly, that each of the ProLiant's processors is up to an Athalon 1.2) and ram of an affordable game box as an enterprise server, more (or, rather, less) power too you.
You've got a *damned* fine CAD machine there. Using it as anything else is a cludge.
We run a couple dozen web apps on a single ultrasparc machine. Four processors, 4gb ram. That's one of the smaller ones. If you had ANY idea what real computers were, you'd know that Windows can't even run on them.
a: Advancd server runs on Intel archetecture systems. It only scales as far as those go, which is high-end workstation.
b: This is Microsoft's own recommended procedure. Don't blame me if they can't keep their own boxen up (Yes, I was at a major EDI service provider that ran their internet app on IIS with SQL Server boxes for the database. Microsoft's only solution to the downtime on the sql server box was "add two redundant servers". To performance on the IIS box, "Load balance across three machines". There was absolutely nothing they could come up with to give the performance and reliability we needed on a single box. Nothing.
c: This was also microsoft recommended procedure, although I admit to it not having been recommended to us (probably because we already had everything on different machines...)
False.
The cost of adminning Windows servers is considerably higher than the cost of adminning Unix servers (of any flavor). You can cover far fewer servers with a single admin, and you need at least 3 times the number of physical boxen than you do with Unix systems because a: Windows scales horribly and only runs on hardware designed to be workstations b: Windows requires at least two redundant servers for each primary server to maintain the uptimes of any Unix c: Windows is only able to perform properly if each box only runs one particular server function. Put a print server, a web server and a file server on the same box and none of them will work well (well by Windows standards).
100 "executives", 60% of whom thought that Windows was going to be their enterprise server of choice. Not their desktop of choice. Not some of their webservers. Their enterprise server.
The poll needed to ask those same executives what they DO use, and correlate that to what actually is used so they could remove answers from people who obviously have no involvement in their company's enterprise server purchaces. My guess is that they answered "Windows" for the same reason I told a telephone survey person that Glitton (however it's spelled) was my exterior paint of choice. It was the only name I could think of at the time, and they just wanted an answer. I answered "Glitton" to every question it was appropriate to.
In the famous Betamax suit, whereby time-shifting was declared legal, the plaintiff trying to restrict Sony's Betamax technology was Sony Entertainment.
In this case, I think we and Microsoft oppose it for pretty much the same reasons. We both want to be able to do whatever we want with our own shit.
Granted, MS certainly tends to try to force others to do what they (ms) want as well, but the only reason that's wrong is that they (like the government) are in the position to succeed in forcing others to comply (or, at least, make it quite painful not to comply).
Or whatever beings or forces of nature you want to thank that the self-interests of major companies and industries are seldom in line with each other. For as long as there's a solid business reason for one industry or company to oppose the actions or legal efforts of another industry or company, those of us who always get caught in the middle will have a chance to survive and maybe even prosper, however we individually understand prospering.
Maybe we'll get really lucky and have Sony Entertainment and Sony Electronics on the opposite sides of another lawsuit. That's always fun.
I would like to introduce the MacGabhain Open Source Award. You may award it to anyone else you like, so long as you don't restrict them from awarding it to others. You may modify the award in any way you like, so long as that award may also be awarded by anyone else to anyone else. You must include the following statement in any issuance of this award:
This award is or includes the MacGabhain Open Source Award. You may grant this award, either in its current form or in any modified form, to anyone provided you allow them to grant this award to anyone else and you include this statement in any granting of the award.
While I found the previous "5 Insightful" rather off too, one would think that in order to be "redundant" a post would actually have to have been made after other posts on the same subject. Having a post at 1:51 following several starting at 1:49 doesn't really seem to qualify for that either. So 3 people modded me up for making the same really obvious observation as 5 other people at the same time, and then three other people modded me down for making the same really obvious observation as 5 other people at the same time.
I'm starting to think that maybe Slashdot deserves to have articles that claim RAM is cheaper per MB than Hard Drives. It might even deserve to have Jon Katz as its feature writer.
No, "per unit of currency" means per dollar or pound or franc or what have you. "Currency" means "money". Ram has *always* been cheaper per physical item - say, $200 for four 1MB 30-pin non-parity vs $450 for a 420MB hard drive in 1993 (when hard drives had been dropping in price for a couple of years while RAM remained stable at around $50/meg).
$20 gets you about 256 MB of ram. $200 gets you about 75,000 MB of HD space. Ten times the price gets you 300 times the MBs. What are you smoking, and can you give some of it to my credit card companies?
Slashdot Mirror
I admit I skimmed most of the article (looking mainly for an answer to this question), but it didn't seem to mention in the system configurations or such which file systems were being used on the Linux and Windows boxes. Since we're talking file server throughput, the file system being used would have a major effect on the "getting the stuff off the drive" part of the test.
Not that this should surprise me.
if some ISP wanted to go after NAT users If your ISP charges per computer (such as cable modem providers), then they should have a broken law on their side. You've contracted to use their service with one machine (and on a shared-bandwidth network, there's good reason for that contract limitation). If you start hooking up more machines, you're effectively splicing those machines onto the network, just like with CableTV. If your ISP authorizes the setting up of networks to share they bandwidth they provide you, the law would not apply.
AMD, on the other hand, is finishing up a 64bit processor that suffers zero (or at worst nominal) slowdowns on systems with 32 bit OSs when compared to comparable 32 bit processors, leaving plenty of room for their power to continue to grow as the software catches up.
There are two factors (other than collusion) that could account for that phenomenon (which I've seen as well, both from the inside and the outside).
First: There are certain products that make sense to advertise because of their drawing power that pretty much every store in the same category is going to use. These are the ones that have nice numbers about them, but are cheap. So, if you've got a particular 17" monitor you can sell for $79.99, you advertise that. This gets people into the store. If they want that monitor, they can, of course, buy it, but you're often hoping that they'll change their mind when they see better options and get something with a better margin. (It's only bait and switch if you don't have the product you advertise.)
Second: Because consumer electronics giants are dumbshits when it comes to pricing (they'd drive themselves out of business by eliminating all margin if they could), the manufactorers place limits on what how cheaply you can advertise their products, in order to make sure their product doesn't become a money-loser and loose support of the store. (Happened to HP ink jets when I worked at CompUSA years ago. Due to Best Buy's use of them as a draw, margins were around $2.50 on them, as opposed to $30 on a Canon. Our store couldn't get HP inkjets in quantity for months.) This is called the Minimum Advertised Price, or MAP. It's semi-legal, since it's an agreement between the store and the manufactorer, not between stores, and because it doesn't prevent you from selling below the price -- only from advertising below the price. So when BB or Circuit City or whomever has that Deskjet 635 on sale for $49.95, they all end up with the same price because they're all discounting it to the MAP.
The sound being produced is not the related to the noise some hard drives make. He's not spinning them at particular rates to get different sounds (which would only allow him to produce one frequency per platter anyway). He's vibrating them, just like regular drivers. Still a cool waste of time, but not the least bit what the posting indicated.
Amid all those "outsiders of the week", ever since the net went public its biggest componants and financial backers have been the huge brick and mortar businesses. Suburban hackers and dot coms have at their best been a far second to IBM, 3M, Merrill Lynch, Wells Fargo, General Motors and all of the other multi-billion dollar companies that, through their own use of the net, covered the costs of building it so the eschewed hackers Jon so loves to think he's in touch with could write a few perl scripts.
If we're talking about ideas that are at the heart of the net and what most represents them, I'd have to say that for the last 8 or 10 years - since the net started to be really popular - that idea has been self-promotion, whether of a mega corporation or an idea or a person. My vote for the epicenter of this use of the net is jennicam.
Dead trees look better on other dead trees then CDs do (much less an ethernet port).
Being a college instructor, I need books on my shelf that people will assume I've read to improve my standing in the tribe.
Having more than two devices on a channel only causes slowdown on a broken channel design. I've had 5 drives chained onto an UW SCSI card and benchmarked better data transfer off of them while simultaneously running the benchmark on all five drives than I got from a singe 7200RPM ATA66 drive in my other system. (This was a couple years back.)
I wouldn't expect most people to realize that, of course. SCSI is, as was noted, really quite pricey. But it's damned fast and doesn't break a sweat being chained. It's unfortunate that it was never able to get into competative price points with IDE and the various kludges that have been made to it over the year. While I don't have the systems to test it, I'd be willing to bet that a fast SCSI II system with the best drives available from 1992 would still blow the doors off of a brand spanking new ATA 100 system in data transfer.
It's CmdrTaco's summary. The article only claims these are "Toys to Avoid", by whatever standards the site uses. Which begs the further question of whether CmndTaco presenting this as "News for Nerds, Stuff that Matters" is misleading as well.
Well, he's obviously old enough to remember a time when he almost had a job as a real journalist. I suspect the scanning idea. Putting in more than the absolute minimum effort needed to get crap published hasn't really been his strong suit.
Any time you see an argument purporting that Unix is easier to admin than WinNT or Novell it's coming from someone who has never spent any time administering real servers.
We've got this nifty little thing here called a SunRay. For those not familiar with them, it's a cool box from Sun that serves up Xterms to dumb terminals. We've got about a dozen terminals attached to it at the moment. While my job is not adminning it, I'm allowed to. I can do absolutely anything to absolutely anyone from any of the terminals, and when I log in, it will be configured just like my desktop, because it is my desktop. They even use these cards that will override the current session with that of the owner of the card as long as the card is inserted in the terminal, so I can go up to someone's machine, stick the card in, and immediately be working on my machine.
Naturally, I can shell into the machine from any other machine in the company, and while I won't have my pretty Xterm, I'll have the same level of control. I can do anything from move files around to fix someone's screen color and resolution.
Of course, this isn't really a good comparison, since MS doesn't even make a multi-user OS, so they don't have anything like a SunRay to take advantage of the fact that, since it's inception, their OS was built to exist as a networked OS.
Any time you see an argument purporting that WinNT/2000 is even in the same league as Unix as a server OS, much less as good or better than Unix, it's coming from someone who's never even seen, or perhaps even conceived of, real servers.
Does your company do web services or any such thing on those servers? Otherwise, how on earth did you manage to amass 70 servers for only 800 users?
You go right ahead and consider your cute little proliant to be a bit more than a high end workstation. My $2000 game machine at home is a dual athalon with 1gb of ram running Win2K. If you want to think of something with only 4 times the processing power (assuming, almost certainly falsly, that each of the ProLiant's processors is up to an Athalon 1.2) and ram of an affordable game box as an enterprise server, more (or, rather, less) power too you.
You've got a *damned* fine CAD machine there. Using it as anything else is a cludge.
We run a couple dozen web apps on a single ultrasparc machine. Four processors, 4gb ram. That's one of the smaller ones. If you had ANY idea what real computers were, you'd know that Windows can't even run on them.
a: Advancd server runs on Intel archetecture systems. It only scales as far as those go, which is high-end workstation.
b: This is Microsoft's own recommended procedure. Don't blame me if they can't keep their own boxen up (Yes, I was at a major EDI service provider that ran their internet app on IIS with SQL Server boxes for the database. Microsoft's only solution to the downtime on the sql server box was "add two redundant servers". To performance on the IIS box, "Load balance across three machines". There was absolutely nothing they could come up with to give the performance and reliability we needed on a single box. Nothing.
c: This was also microsoft recommended procedure, although I admit to it not having been recommended to us (probably because we already had everything on different machines...)
False.
The cost of adminning Windows servers is considerably higher than the cost of adminning Unix servers (of any flavor). You can cover far fewer servers with a single admin, and you need at least 3 times the number of physical boxen than you do with Unix systems because a: Windows scales horribly and only runs on hardware designed to be workstations b: Windows requires at least two redundant servers for each primary server to maintain the uptimes of any Unix c: Windows is only able to perform properly if each box only runs one particular server function. Put a print server, a web server and a file server on the same box and none of them will work well (well by Windows standards).
100 "executives", 60% of whom thought that Windows was going to be their enterprise server of choice. Not their desktop of choice. Not some of their webservers. Their enterprise server.
The poll needed to ask those same executives what they DO use, and correlate that to what actually is used so they could remove answers from people who obviously have no involvement in their company's enterprise server purchaces. My guess is that they answered "Windows" for the same reason I told a telephone survey person that Glitton (however it's spelled) was my exterior paint of choice. It was the only name I could think of at the time, and they just wanted an answer. I answered "Glitton" to every question it was appropriate to.
Yeh. It's called C++.
In the famous Betamax suit, whereby time-shifting was declared legal, the plaintiff trying to restrict Sony's Betamax technology was Sony Entertainment.
In this case, I think we and Microsoft oppose it for pretty much the same reasons. We both want to be able to do whatever we want with our own shit.
Granted, MS certainly tends to try to force others to do what they (ms) want as well, but the only reason that's wrong is that they (like the government) are in the position to succeed in forcing others to comply (or, at least, make it quite painful not to comply).
Or whatever beings or forces of nature you want to thank that the self-interests of major companies and industries are seldom in line with each other. For as long as there's a solid business reason for one industry or company to oppose the actions or legal efforts of another industry or company, those of us who always get caught in the middle will have a chance to survive and maybe even prosper, however we individually understand prospering.
Maybe we'll get really lucky and have Sony Entertainment and Sony Electronics on the opposite sides of another lawsuit. That's always fun.
I would like to introduce the MacGabhain Open Source Award. You may award it to anyone else you like, so long as you don't restrict them from awarding it to others. You may modify the award in any way you like, so long as that award may also be awarded by anyone else to anyone else. You must include the following statement in any issuance of this award:
This award is or includes the MacGabhain Open Source Award. You may grant this award, either in its current form or in any modified form, to anyone provided you allow them to grant this award to anyone else and you include this statement in any granting of the award.
While I found the previous "5 Insightful" rather off too, one would think that in order to be "redundant" a post would actually have to have been made after other posts on the same subject. Having a post at 1:51 following several starting at 1:49 doesn't really seem to qualify for that either. So 3 people modded me up for making the same really obvious observation as 5 other people at the same time, and then three other people modded me down for making the same really obvious observation as 5 other people at the same time.
I'm starting to think that maybe Slashdot deserves to have articles that claim RAM is cheaper per MB than Hard Drives. It might even deserve to have Jon Katz as its feature writer.
No, "per unit of currency" means per dollar or pound or franc or what have you. "Currency" means "money". Ram has *always* been cheaper per physical item - say, $200 for four 1MB 30-pin non-parity vs $450 for a 420MB hard drive in 1993 (when hard drives had been dropping in price for a couple of years while RAM remained stable at around $50/meg).
$20 gets you about 256 MB of ram. $200 gets you about 75,000 MB of HD space. Ten times the price gets you 300 times the MBs. What are you smoking, and can you give some of it to my credit card companies?
Sheesh. So they want their whole department to grind to a halt instead of just one cruiser?