This problem has been a forfront issue for me in my job for the last couple of years, so understand that nothing I have to say on it is based on speculation, or an article in a tech mag, it's from the real world.
Heat sinks, air flow, etc does NOT reduce the thermal load of a datacenter, it only improves the efficiency of a single server dumping it's waste heat into the space. THAT is the problem to deal with. Personally I can't afford to give a rats ass about one box's cooling, if there's over a thousand more that all have the same needs. If it doesn't help the datacenter as a whole, it's just trivia.
APC makes some cool air handling cabinets that are sealed and vented to draw in cool air, and vent the hot directly into ducting. very slick, but still that does NOT change the thermal loading of the datacenter.
What to do with the waste heat is the #1 problem, and really the only problem in this discussion. fans and heat sinks are fine to get the waste heat out of the servers, but it has to go someplace.
In a traditional HVAC CRAC environment, hot air is chilled in the unit, waste heat is exchanged into a water supply, which flows out of the room to a chiller unit that pumps the heat out into the external air.
A very interesting alternative I've seen gaining popularity is to use ambient cooling. What that means is if you want to keep your datacenter below "x" degrees, then any time the outside ambient air tempature is below that, you simply open a window. It's a little more complicated, but that's the gist, draw in outside air that's cooler, blow out the hot air. In some locations (Seattle for example) the chillers actually only have to run less than 1% of the time. It's obviously less effective in hot locations, but eliminating the cycle of re-cooling your own hot air is a huge win both for cost and efficiency.
The other alternative is better air handling (ducting, raised floor, etc) combined with better cooling, combined with more chiller water supply. The latter is typically the bottleneck.
water cooling in the datacenter only makes sense IF the heated water is vented to outside the datacenter. A central piping system that goes through the heatsink on each server, and vents outside would be great, but complex and dangerous far beyond the benefits. Water cooling that is inedepentant on each server is just a more efficient way to get waste heat out into the datacenter, and hasn't solved anything.
I like jokes as much, probably more, than most people. Usually, in as much poor taste as possible.
However, where I draw the line is when journalistic, or supposedly credible information resources resort to disseminating false information under the guise of "good fun"
It's especially bad in magazines where you might be reading the April 1 issue a month or more before the actual date. PC Magazine was somewhat notorious for this, and eventually I found myself wondering if some articles were real or just BS.
Is that REALLY the position a tech resource wants it's readers to be in?
Leave the jokes and tricks to friends, coworkers, and email, keep it away from anyplace you expect to be taken seriously.
I agree that that's "probably" what he did, but if you're going to post benchmarks, they're kind of useless without the accompanying testing criteria.
For example, in addition to the possibility I listed before, rundown tests could be very different if one was word processing tasks, while the other was watching a DVD.
Battery benchmarking is very difficult for this reason, power management features can make a huge difference without a necessarily level playing field.
The first one would have used the combined juice from his internal laptop battery and the external one, but the second one would have started with an already depleted internal battery.
Unless he left out a lot of detail (like: I brought along a second fully charged internal battery, and swapped it out in the middle of the flight...) the rundown test on the second battery is completely invalid.
I keep having to point this out for some reason: deliberately moving the swapfile to the outer sectors will take advantage of the faster I/O from that location (never heard anyone suggest the middle for "convenience" before though...hah). However, that's also got the effect of deliberately moving your data and OS to the slower parts of the disk.
This is not an optomization scheme, this is a misguided overthinking of a simple concept. Swapping memory to disk is a necessary evil that needs to be avoided at all costs if you're concerned about performance. Use a seperate disk for swap, or buy more ram. anything else is an exercise in futility.
My front page is 83k of HTML, 10k of CSS, 20-50k of images, and, in one part, 140k of javascript.
Frankly speaking dude, a person who calls themselves a "web developer" and is making 283K homepages is part of the problem. That's bigger than CNN's.
Badmouthing people for your inability to control your page bloat, just shows that your maturity as a developer is lacking in more areas than just efficiency.
I thought the first partition, being closer to the inside of the platter, had shorter seek times. This means that the arm would move less hence less wear and tear. Also means faster to access information than that stored on the outside of the platter hence better performance. Can you point out the site that proves this to be urban myth?
Phillip.
Phillip,
You should really read the article before thinking you understand the commentary on it. Let me help you out. The statement was:
You should always have a dedicated partition for your temp files and swap file. It's tempting to actually put this on a separate physical drive to reduce the wear and tear on the main drive, but the disadvantage is that upgrading to a larger hard drive a more involved process.
Note that it says "dedicated partition", not "first partition".
If one were actually to go to the trouble of installing XP with the intention of using the first partition as swap, and second etc for OS and data, there would be some validity to the idea. However, you'd be putting your OS and data deliberately on slower parts of the disk, to speed up the swapfile which you don't want to ever use anyway (if you can help it).
Just trust me, if you're performance tunig your swapfile, you're putting your efforts into the wrong bucket. You need more ram.
Considering that you acknowledge that the article and discussin were about XP, not Linux, I find it odd that you'd jump into the middle of it claiming insult to Linux programmers.
The article shows it's roots in voodoo and urban myth with the statement about moving the swap file to a separate partition.
A separate partition is STILL THE SAME DRIVE. Same platters, same heads. The only benefit is that it's a little cleaner to look at.
If you need better swap performance, the ONLY way to get it is to move the swapfile to a seperate, hopefully faster, drive.
However, if you're looking for ways to improve your swapfile performance, you're a freakin' idiot who needs to stop touching PC's.
Swapfile is a necessary evil, if swapping is degrading your performance YOU NEED MORE RAM, not a faster swapfile. It's not rocket science. That $150 you'd spend on a dedicated swap drive would buy you a gigabyle of RAM and end the problem forever.
sure, you could reduce wear and tear on your OS drive by splitting the data out across as many drives as you can stick in there.
However, due to the realities of MTBF, every drive you add increases your chances of a catastrophic failure. If you don't have a real performance reason for adding spindles, more drives is just more points of failure.
You listed 7 bullet points that is really only 3 points.
1, 5, and 7 are the same thing. you listed itunes 3 times as if nobody would notice, and is a dubious opinion not a fact.
2 is utter bullshit that you could only state as a deliberate falsehood, or to just be unaware that other people than Apple make MP3 players. Apple makes some of the most EXPENSIVE Mp3 players, not cheapest.
3 and 4 are the same thing.
and 6 is a repeat of the problem with 2. What the hell is different about the interface on the Ipod than ANY other mp3 player on the market? It's got a stop, play, ff etc button. I'm suprised you didn't accuse the rest of the world of copying apple for using the same set of buttons.
I personally think Apple makes good products, and I welcome the competition to the portable game because it drives the prices down and quality up for everyone. However, there's really no arguing that apple joined a market 5 years late and capitalized by stealing the ideas that everyone else had been marketing. Extremely minor changes do not make it an innovative product.
well, I'm probably within your "old fart" definition range, but it wasn't so long ago that I don't remember seeing Star Wars for the 20th or so time and suddenly saying "Episode WTF???"
Those of us old enough to actually remember will tell you Lucas originally promised both Jack and Shit.
The whole "Episode IV" thing didn't even come about until the theatrical re-release of Star Wars about 18 months after the initial one.
Lucas likes to pretend he's had this grand vision all along from day one, but the plot inconsistancies and herky-jerky flow of the story looks more like incoherent post-facto ramblings than it does "planned".
Which is exactly why it would be inappropriate to use it as a single factor authentication. combining it with one, or two even better, other factors helps guarantee that if the security isn't unbreakable, it's as impractical to do so as possible.
the typical rule of thumb that gets erpeated ad nauseum is "something you know plus something you have or something you are".
examples of "know" would be a password or pin number
something you have is like a key card, or an RSA keyfob with a rotating code number
something you are is a fingerprint, a retina pattern, palm, face match, etc.
So the best solution would be a PIN/Password, plus an RSA keyfob, plus fingerprint. That would be next to impossible to compromise without the cooperation of the person.
This is the kind of thinking I have to fight every day at work. A simple lack of understanding of the concept makes a useless solution seem perfectly reasonable. I don't mean to be as insulting as that sounds, this is just a good example of how easy it is to be completely wrong.
If you start with a known item like the time (time changes, but it's not a secret what time it is) then multiply it by another unchanging item like a PIN, all you've done is make a more complicated PIN number. You haven't implemented two factor authentication, you're just making it hard to log in.
What exactly is wrong with so called journalists? Apple releases a "me too" product about 4 years late to the game and Sony gets accused of copying them? Sony has made flash based players like this for half a freakin' decade.
It's like "If Apple doesn't make one, it doesn't exist. But if Apple decides to make one, then it's better than anything else ever. Oh, and Apple invented everything, even things that existed for years before they entered the market".
As any Apple-watcher will tell you, this has been SOP with the Wintel world for decades now.
And as any industry-watcher will tell you, this is the oft repeated mantra of the Apple fanatic that ignores every blatent rippoff and blunder that Apple commits.
Let's see if I can pull a few things off the cuff that Apple has copied from the "PC" industry...
1. USB
2. PCI Bus
3. IDE drives
4. Everything remotely related to the iPod
5. Apple mini form factor (Intel's announcement aside, smaller PC's than this have been around for YEARS).
but I forget, Apple is allowed to copy anything they like without any comment from you idiots.
how about this: how about we just accept that building upon the ideas of your competition and one-upping them is what has spurred the personal computer industry to grow faster than any in history. Of course that would require you to quit your childish bitching...then where would you be?
Has the PC industry really gotten that bad so that they don't do anything but copy Apple? First eMachines copies the iMac, now Intel is trying to show that PCs can be mini too. I know it's a mockup, but do something original instead of copying the color of the mini.
It's never too difficult to pick up on where people's agendas lie. Calling this a "copy of apple" can only be explained by ignoring the fact that sub-micro form factors have been available for YEARS in the PC world.
It would be more accurate to suggest that lately Apple has had a lot of success copying things from OTHERS. Contrary to popular belief the iPod was in no way original, but a very late to the game "me too" product, and the Apple Mini is quite a bit LARGER than several PC's that have been around for years. The "strikingly similar" comment is utterly laughable, unless Apple has managed to copyright the "square" shape now.
Love Apple all you like, but when you twist reality to support your point of view, you look pretty damned ignorant.
the gyros don't keep it in orbit, they keep it stabilized. All the gyros could stop and it wouldn't have an appreciable effect on the orbit decay.
Booster rockets are used to refresh the orbit.
On Christmas night, two SkyWest pilots said they saw two laser-like rays of light in their cockpit as they attempted to land at the airport in Medford, Ore.
On Monday, a laser beam was directed into the cockpit of a commercial jet flying about 15 miles from Cleveland Hopkins International Airport at an altitude of between 8,500 and 10,000 feet, the FBI said. It was determined the laser came from a residential area in suburban Warrensville Heights.
Also on Monday in Colorado Springs, two pilots reported green pulsating laser lights beamed into their cockpits. Police sent patrol cars and a helicopter in a fruitless search.
In New Jersey, the pilot of a corporate-owned Cessna Citation carrying 13 people said three green lasers were pointed into his cockpit while approaching the Teterboro airport on Wednesday night. Law enforcement officials said they were believed to have originated near a mall in Wayne.
I'm sorry, but don't these anecdotes remind you just a little bit of the "mysterious odor sends dozens to hospital at local mall" stories on the news?
One person says they smell a laser, someone else says "hey, I smell lasers in my cockpit too". Next thing you know, you're nobody unless terrorists are illuminating your cockpit with lasers. Lasers that smell mysterious.
This almost exact idea was played out in a Tom Clancy novel, I think it was "Debt of Honor". One of the CIA guys (or whatever agency it was, it was a long time ago) used an ultra-high powered spotlight to blind the pilots of landing airliners, causing them to crash (read the book, it explains why he was still a good guy).
I could probably make a pretty good show of it with my handheld, battery powered 2.5 million candlepower spotlight out at the local airport. Or for that matter, on the local interstate (If I'm satisfied with carnage on a smaller scale).
I think that the 'green dot' would have to be from a laser which was up there with the airliner.
Exactly my thought. I'll bet the copilot was screwing with the pilot, and now is afraid to admit it cuz he'd be in it deep.
As far as the military goes, I don't know much about weapons guidance systems, but I'll bet they're not VISIBLE lasers.
Slashdot Mirror
Heat sinks, air flow, etc does NOT reduce the thermal load of a datacenter, it only improves the efficiency of a single server dumping it's waste heat into the space. THAT is the problem to deal with. Personally I can't afford to give a rats ass about one box's cooling, if there's over a thousand more that all have the same needs. If it doesn't help the datacenter as a whole, it's just trivia.
APC makes some cool air handling cabinets that are sealed and vented to draw in cool air, and vent the hot directly into ducting. very slick, but still that does NOT change the thermal loading of the datacenter.
What to do with the waste heat is the #1 problem, and really the only problem in this discussion. fans and heat sinks are fine to get the waste heat out of the servers, but it has to go someplace.
In a traditional HVAC CRAC environment, hot air is chilled in the unit, waste heat is exchanged into a water supply, which flows out of the room to a chiller unit that pumps the heat out into the external air.
A very interesting alternative I've seen gaining popularity is to use ambient cooling. What that means is if you want to keep your datacenter below "x" degrees, then any time the outside ambient air tempature is below that, you simply open a window. It's a little more complicated, but that's the gist, draw in outside air that's cooler, blow out the hot air. In some locations (Seattle for example) the chillers actually only have to run less than 1% of the time. It's obviously less effective in hot locations, but eliminating the cycle of re-cooling your own hot air is a huge win both for cost and efficiency.
The other alternative is better air handling (ducting, raised floor, etc) combined with better cooling, combined with more chiller water supply. The latter is typically the bottleneck.
water cooling in the datacenter only makes sense IF the heated water is vented to outside the datacenter. A central piping system that goes through the heatsink on each server, and vents outside would be great, but complex and dangerous far beyond the benefits. Water cooling that is inedepentant on each server is just a more efficient way to get waste heat out into the datacenter, and hasn't solved anything.
However, where I draw the line is when journalistic, or supposedly credible information resources resort to disseminating false information under the guise of "good fun"
It's especially bad in magazines where you might be reading the April 1 issue a month or more before the actual date. PC Magazine was somewhat notorious for this, and eventually I found myself wondering if some articles were real or just BS.
Is that REALLY the position a tech resource wants it's readers to be in?
Leave the jokes and tricks to friends, coworkers, and email, keep it away from anyplace you expect to be taken seriously.
For example, in addition to the possibility I listed before, rundown tests could be very different if one was word processing tasks, while the other was watching a DVD.
Battery benchmarking is very difficult for this reason, power management features can make a huge difference without a necessarily level playing field.
Overall, I trust his opinion though.
The first one would have used the combined juice from his internal laptop battery and the external one, but the second one would have started with an already depleted internal battery.
Unless he left out a lot of detail (like: I brought along a second fully charged internal battery, and swapped it out in the middle of the flight...) the rundown test on the second battery is completely invalid.
This is not an optomization scheme, this is a misguided overthinking of a simple concept. Swapping memory to disk is a necessary evil that needs to be avoided at all costs if you're concerned about performance. Use a seperate disk for swap, or buy more ram. anything else is an exercise in futility.
now THAT makes for an awkward front page.
Frankly speaking dude, a person who calls themselves a "web developer" and is making 283K homepages is part of the problem. That's bigger than CNN's.
Badmouthing people for your inability to control your page bloat, just shows that your maturity as a developer is lacking in more areas than just efficiency.
You should really read the article before thinking you understand the commentary on it. Let me help you out. The statement was:
You should always have a dedicated partition for your temp files and swap file. It's tempting to actually put this on a separate physical drive to reduce the wear and tear on the main drive, but the disadvantage is that upgrading to a larger hard drive a more involved process.
Note that it says "dedicated partition", not "first partition".
The outer sectors are the faster ones, not the inner ones, although your assumption about that being where the first partition ends up does hold true. Here's a good explaination of why: http://www.dewassoc.com/kbase/hard_drives/hard_dis k_sector_structures.htm
If one were actually to go to the trouble of installing XP with the intention of using the first partition as swap, and second etc for OS and data, there would be some validity to the idea. However, you'd be putting your OS and data deliberately on slower parts of the disk, to speed up the swapfile which you don't want to ever use anyway (if you can help it).
Just trust me, if you're performance tunig your swapfile, you're putting your efforts into the wrong bucket. You need more ram.
Lighten up...or at least try to stay on subject.
A separate partition is STILL THE SAME DRIVE. Same platters, same heads. The only benefit is that it's a little cleaner to look at.
If you need better swap performance, the ONLY way to get it is to move the swapfile to a seperate, hopefully faster, drive.
However, if you're looking for ways to improve your swapfile performance, you're a freakin' idiot who needs to stop touching PC's.
Swapfile is a necessary evil, if swapping is degrading your performance YOU NEED MORE RAM, not a faster swapfile. It's not rocket science. That $150 you'd spend on a dedicated swap drive would buy you a gigabyle of RAM and end the problem forever.
I guess anyone can write an article...
However, due to the realities of MTBF, every drive you add increases your chances of a catastrophic failure. If you don't have a real performance reason for adding spindles, more drives is just more points of failure.
1, 5, and 7 are the same thing. you listed itunes 3 times as if nobody would notice, and is a dubious opinion not a fact.
2 is utter bullshit that you could only state as a deliberate falsehood, or to just be unaware that other people than Apple make MP3 players. Apple makes some of the most EXPENSIVE Mp3 players, not cheapest.
3 and 4 are the same thing.
and 6 is a repeat of the problem with 2. What the hell is different about the interface on the Ipod than ANY other mp3 player on the market? It's got a stop, play, ff etc button. I'm suprised you didn't accuse the rest of the world of copying apple for using the same set of buttons.
I personally think Apple makes good products, and I welcome the competition to the portable game because it drives the prices down and quality up for everyone. However, there's really no arguing that apple joined a market 5 years late and capitalized by stealing the ideas that everyone else had been marketing. Extremely minor changes do not make it an innovative product.
well, I'm probably within your "old fart" definition range, but it wasn't so long ago that I don't remember seeing Star Wars for the 20th or so time and suddenly saying "Episode WTF???"
The whole "Episode IV" thing didn't even come about until the theatrical re-release of Star Wars about 18 months after the initial one.
Lucas likes to pretend he's had this grand vision all along from day one, but the plot inconsistancies and herky-jerky flow of the story looks more like incoherent post-facto ramblings than it does "planned".
Which is exactly why it would be inappropriate to use it as a single factor authentication. combining it with one, or two even better, other factors helps guarantee that if the security isn't unbreakable, it's as impractical to do so as possible.
the typical rule of thumb that gets erpeated ad nauseum is "something you know plus something you have or something you are".
examples of "know" would be a password or pin number
something you have is like a key card, or an RSA keyfob with a rotating code number
something you are is a fingerprint, a retina pattern, palm, face match, etc.
So the best solution would be a PIN/Password, plus an RSA keyfob, plus fingerprint. That would be next to impossible to compromise without the cooperation of the person.
If you start with a known item like the time (time changes, but it's not a secret what time it is) then multiply it by another unchanging item like a PIN, all you've done is make a more complicated PIN number. You haven't implemented two factor authentication, you're just making it hard to log in.
It's like "If Apple doesn't make one, it doesn't exist. But if Apple decides to make one, then it's better than anything else ever. Oh, and Apple invented everything, even things that existed for years before they entered the market".
And as any industry-watcher will tell you, this is the oft repeated mantra of the Apple fanatic that ignores every blatent rippoff and blunder that Apple commits.
Let's see if I can pull a few things off the cuff that Apple has copied from the "PC" industry...
1. USB
2. PCI Bus
3. IDE drives
4. Everything remotely related to the iPod
5. Apple mini form factor (Intel's announcement aside, smaller PC's than this have been around for YEARS).
but I forget, Apple is allowed to copy anything they like without any comment from you idiots.
how about this: how about we just accept that building upon the ideas of your competition and one-upping them is what has spurred the personal computer industry to grow faster than any in history. Of course that would require you to quit your childish bitching...then where would you be?
It's never too difficult to pick up on where people's agendas lie. Calling this a "copy of apple" can only be explained by ignoring the fact that sub-micro form factors have been available for YEARS in the PC world.
It would be more accurate to suggest that lately Apple has had a lot of success copying things from OTHERS. Contrary to popular belief the iPod was in no way original, but a very late to the game "me too" product, and the Apple Mini is quite a bit LARGER than several PC's that have been around for years. The "strikingly similar" comment is utterly laughable, unless Apple has managed to copyright the "square" shape now.
Love Apple all you like, but when you twist reality to support your point of view, you look pretty damned ignorant.
the gyros don't keep it in orbit, they keep it stabilized. All the gyros could stop and it wouldn't have an appreciable effect on the orbit decay. Booster rockets are used to refresh the orbit.
Yeah, that's right. I couldn't remember why it was ok to crash passenger aircraft, now it makes sense.
;-)
Like I said, I read it a looong time ago.
I'm sorry, but don't these anecdotes remind you just a little bit of the "mysterious odor sends dozens to hospital at local mall" stories on the news?
One person says they smell a laser, someone else says "hey, I smell lasers in my cockpit too". Next thing you know, you're nobody unless terrorists are illuminating your cockpit with lasers. Lasers that smell mysterious.
This almost exact idea was played out in a Tom Clancy novel, I think it was "Debt of Honor". One of the CIA guys (or whatever agency it was, it was a long time ago) used an ultra-high powered spotlight to blind the pilots of landing airliners, causing them to crash (read the book, it explains why he was still a good guy).
I could probably make a pretty good show of it with my handheld, battery powered 2.5 million candlepower spotlight out at the local airport. Or for that matter, on the local interstate (If I'm satisfied with carnage on a smaller scale).