And thank $deity IBM finds Linux useful in making them money. IBM's contributions to Linux have been phenomenal, not just people, and code but dollars as well. I don't know what figure, but I'd guess IBM's $$ contribution to OSDL et al is sizable.
I have worked in disk storage design. This was a very cool project. This looks like a promising start and in some ways represents the future of storage; COTS parts. Others have pointed out some areas of improvement, cooling and the like.
These guys will either become a Nexsan or NetAPP. Or, they'll starve and die. Anyone, _anyone_ can build out a huge set of cots parts, sell it, and leave the management burden to the customer. Unless this customer is a U.S. nuclear weapons lab or university environment with a ton of essentially cheap doctoral candidate talent and 'free' man-time, this kind of solution fails every time because it's not manageable at scale. Once you add the management capability to the hardware and write the management software interface, you ARE the Nexsan of about 10 years ago. Currently, Nexsan is the closest thing you can get to "cots" prices while still getting some value add capability and single source warranty contact.
Depends on what kind of server. If you're talking about a Web server, IIS 5.1 and later or Apache 2.x and better with multithreading on, yes. If you're talking about Apache 1.x or 2.x without multithreading, or some older versions of IIS, no.
Apache 1.x forks child processes. AFAIK, 2.x does the same. On *nix, any process can be scheduled on any CPU in the system. So your argument doesn't hold water. *nix has been inherently "multi-threaded" or "multi-tasking" for a very long time, on the order of 30 years.
If you're arguing about Apache Win32 and how Windows handles processes and threads, then it doesn't matter, as Windows sucks as a dedicated server platform, web or otherwise, and isn't worth discussion.
Should have put this in my first post. The thermal conductivity of copper is 401 W/mK, concrete is 0.42 W/mK. This gives us a ratio of approx 950:1, close to my guess of 1000:1. This according to: http://www.engineeringtoolbox.com/thermal-conductivity-d_429.html
Concrete is 950 times less efficient than copper. I don't think the author has enough pipe. He may end up adding a small rad and fan in the basement on either the input or output pipe, doesn't really matter as it's a closed loop. Of course, this is all dependent on his PC, and he didn't state if it's a monster gaming system with SLI and a huge overclock or just an average browser/email box.
Talk about a heat sink. I'm a little surprised that this sort of technique is not more widely adopted at places like data centers
Because concrete has horrible thermal conductivity. Its thermal resistance is orders or magnitude greater than a copper CPU heatsink (without a fan)--we're talking around 1000 fold. Given this gigantic thermal resistance, six meters (20 feet) will likely not be enough to dissipate the heat load from the CPU and/or GPU. The starting, or at rest, temperature of the concrete (61F) is irrelevant. What's critical is how quickly it can absorb the heat out of the piping. If it can't absorb it fast enough, the water temperature will continue to rise, eventually thermaling the CPU/GPU.
Lets say the heat the PC is dumping into the water loop is 100W. If the concrete can only absorb 0.001W/cm^2/sec (possibly a high estimate) from the copper pipe, he'll need 10,000 cm2 of pipe surface area, or 10.76 ft^2, if my math is correct. If he's using standard 1/2" copper plumbing pipe, his surface area for 6 meters is approximately 2.62 ft^2, less than 1/3rd of the approx surface area needed. The thermal resistance of the concrete is dependent on the mix the contractor pours. I have no idea what the exact thermal resistance of the author's concrete will be, but the number with be astronomically high compared to numbers we're all familiar with for copper and aluminum.
Something the author didn't think about, apparently, is that concrete cures via a chemical process that actually _generates_ heat until it is cured. So he'd best plan on not firing up his water loop until after the foundation is fully cured. If you recall, they cast such pipes into the Hoover dam (or was it the Glen Canyon damn?) to do exactly the opposite of what the author is attempting. They ran a water loop through a multi thousand ton freon chiller setup to suck the heat out and cut the curing time of the concrete from 100 years (natural cure time) to just a year or so. Granted that was a few million yards^3 of concrete, but you get the point.
I hope it works out. My gut instinct and rough calculations say 20 ft of pipe won't be enough, unless maybe it's 3/4" or 1" pipe in the slab. I hope we see a future article with results.
I'd like you to hand in your geek card, real geeks read the comic books, THEN saw the movies.
I was aiming for the low hanging fruit Zelucifer. If s/he hasn't seen the movies, s/he def hasn't heard of (or read) the comics. However, given pop culture media and advertising being _everywhere_, s/he must have at least heard of the movies.
You've made an incorrect assumption about me because you didn't comprehend my initial post.
...and eventually physically show up at your door, what then?
I'd live for that day.
As long as the dead body is inside your door transom, it's self defense. Just make sure you replace the calculator in his/her hand with an ice pick, knife, of pistol. The police will more readily accept your self defense claim with a weapon in the dead hand of the 4'9" slight build brunette female on the floor. Trust me on this.
And when they keep calling you at your job, and insist on speaking to your boss, and call your family, and eventually physically show up at your door, what then?
They may not use the Linux kernel, but OS X includes a metric shedload...
I believe the quantity you were searching for is 'metric fuckload'. One metric fuckload is greater than a metric shedload by a factor of 10. And there are at least two metric fuckloads of GNU code running around all over the *nix landscape.
Parents, see what happens when you send your kids to public schools? They can't even learn fucking quantities...;)
B&W use it for their tweeters in their high end speakers, as an example.
They're merely coating an aluminum or titanium diaphragm with diamond dust though, not making diamonds. IIRC they use a super high tech press and mold. The press generates something insane like 100,000 PSI, which fuses the diamond shards to each other and the aluminum diaphragm, creating a smooth solid diamond surface on the concave side of the diaphragm. The result is an extremely light yet rigid diaphragm that doesn't color the high frequencies as much as lesser technologies.
Article contains the terms "ATM Machine" and "PIN Number". Read at your own risk.
People - and by this I mean people on Slashdot, I've not seen anyone complain about it elsewhere - always complain about that. But what's the alternative?
It could be referred as "Personal Identification Number" which is just overly long and besides, everybody just knows it as PIN. They could just say "it would scan their card information and record the PINs they entered" but I don't think it is very good. I know the capitalization makes the necessary difference between "pins" and "PINs" here but honestly, that version still looks a bit out of place to me.
One could say "PIN code". It is the version usually used here in Finland ("PIN-koodi") but the difference to PIN number gets very small.
PIN isn't just an acronym for Personal Identification Number. It is, in itself, a name for a short, usually 4 to 8 digits long digit based password. I could bet a lot of money that most of people don't convert the acronym to words when they read text.
Besides, the ATM machine is used what, once? Most of the time it uses just ATM.
With the massive amount of acronyms we have, especially short ones, a lot of them have multiple meanings. While it is relatively easy to understand these ones in this context, I fully support people adding an additional word to tell which meaning of some acronym is meant in a given situation. At least once in an article. There has been too many times I've seen some acronym, tried to google it, found a dozen different meanings and have had no idea of which it refers to.
This is covered in 9th grade English composition class here in the U.S.--or at least it was when I was in 9th grade.
At the beginning of one's literary piece/article, one introduces the full title of a term to be abbreviated by acronym later in the piece. In other words, you spell it out the first time you make reference to the term at the top of the work/article, then you use the acronym in the rest of the work. Anyone sitting in front of a PC should know this. It's a farking world wide literary composition standard.
According to TFA, he's not comparing apples to apples. He doesn't state that he verified each burn after the burn, 6-7-8-9 years ago. Thus, there is no way to know whether the bad discs were bad 9 years ago (i.e. a bad burn), or went bad over time. Given the physics and materials involved, I'd say all his errors today were immediately noticeable after the initial burn. He just didn't bother to check. His methodology isn't even close to scientific, anecdotal at beast. This kid is merely yelling "fire" in the CD-R theater.
Ahh, ok. Now we're on the same page. TFA is about vehicle fuel, and thus I thought you were stating the 'locomotive' fuel of the airship was hydrogen. You're absolutely correct regarding hydrogen being what 'fueled' the catastrophic fire.
I think you've been caught in that situation where you're talking about something you really don't know enough about. Some folks call that bullshitting. Thus, I call bullshit.
Sounds like a misconfiguration. Some dnsbls explicitly state NOT to use them for URL scoring or blocking due to this very reason. AFAIK, there are only a handful of dnsbls worldwide that can actually be used this way properly.
Whether it's a fair world or not, blacklisting entire blocks and not just the bot-infected or spammer hosts does more harm than good, especially when you're talking about blocking entire netblocks that cross multiple businesses full of non-spammer customers.
Unfortunately one of the roles a dnsbl plays in society is carrying a big stick. Listing large'ish blocks makes legit customers complain to their host. If the host doesn't eject the bad apples from their network, they may lose all those legit customers. Yes, unfortunately, the innocent bleed in the process as well.
Back in the day, SPEWS was effective at this, and network operators were forced to be vigilant in keeping spammers off their nets. Nearly everyone hated SPEWS, but, it was effective. Aggressive dnsbls don't fight as much to keep spam out of inboxen as they fight to get spammers booted from their host.
Stopping spammers is not a technical problem, and cannot be accomplished by technical means. It is a social problem, and business problem, and can only be solved by stopping the spammers from being able to send their spew--preempted at the source. The only way to do this is to keep them off the networks. Legislation is also useless. dnsbls are a critical component of this effort. However, not all dnsbls are equal, and their criteria for listing can be very different.
When you need to scream at someone, scream first at the receiver who rejected your mail. HE rejected it, not the dnsbl. The dnsbl is a tool only and provides data. The receiving mail admin makes the decision to reject an email based on that data. If that admin tells you "buzz off spammer", then scream to your provider to get the real spammers off the network. In other words, YOU have to spend some time doing research in the fight, not just screaming that it's someone else's problem that your email was blocked. The fight against spam includes us ALL. It's going to waste some of your time. Blame this on the spammers, not the dnsbls.
My brother's domain got randomly blacklisted, as did another business venture I'm involved with.
dnsbls are databases of IP addresses. They do not contain domains (example.tld). Thus, your brother's domain did not get black listed. And, contrary to your ignorance, there is no 'randomness' to a block list entry. Each IP address that goes into the database goes there for a very specific reason. Usually the reason is spam sent to a spam trap address from said IP address.
Care to be specific around your actual circumstances.
I get the odd call from a Postini client who's been on mxtoolbox.com, crying, "why am I blacklisted? zomg!". SORBS == idiots.
Really? Idiots? I've never received a legit email from Postini, but I've received tons of spam from them. I'd say in this case SORBS is on the money.
lusers and spammers seem to congregate at the same watering holes, intermixed with one another: Postini, Yahoo, Hotmail, AOL, etc. When the only weapon you have is a shotgun, you're bound to hit some lusers even though you're aiming at the spammers.
filter *URL's pointing to a PBL'd IP that are embedded in a message*!!!
My university does that, too. I run a student organization site that has a university subdomain, but is hosted on a shared host. The host inexplicably got listed in the CBL several times, and that screwed up email for the organization staff, and mailing lists for hundreds of students for days at a time.
I didn't realize anyone else used this brilliant filtering scheme.
So, you admit it's a shared server. It gets listed. So, instead of actually finding out what caused the problem, likely a bot infection, you blame the entity which alerted you to said problem.
When you take your car to the shop, and the mechanic finds something wrong with the car that you didn't already know about, do you then blame the mechanic for the problem?
The primary function of dnsbls is to help prevent spam from reaching inboxen. A very important secondary function of dnsbls is to alert good citizens to the fact that their host is spewing, whatever the reason. It's up the the owner of the listed host to dig in and find out what's wrong. Too many people are too fucking lazy to bother these days. So, they just blame the dnsbl as the source of the problem, when in fact, the dnsbl is merely delivering the message that something is wrong.
Slashdot Mirror
IBM is on IBM's side.
And thank $deity IBM finds Linux useful in making them money. IBM's contributions to Linux have been phenomenal, not just people, and code but dollars as well. I don't know what figure, but I'd guess IBM's $$ contribution to OSDL et al is sizable.
I have worked in disk storage design. This was a very cool project. This looks like a promising start and in some ways represents the future of storage; COTS parts. Others have pointed out some areas of improvement, cooling and the like.
These guys will either become a Nexsan or NetAPP. Or, they'll starve and die. Anyone, _anyone_ can build out a huge set of cots parts, sell it, and leave the management burden to the customer. Unless this customer is a U.S. nuclear weapons lab or university environment with a ton of essentially cheap doctoral candidate talent and 'free' man-time, this kind of solution fails every time because it's not manageable at scale. Once you add the management capability to the hardware and write the management software interface, you ARE the Nexsan of about 10 years ago. Currently, Nexsan is the closest thing you can get to "cots" prices while still getting some value add capability and single source warranty contact.
Depends on what kind of server. If you're talking about a Web server, IIS 5.1 and later or Apache 2.x and better with multithreading on, yes. If you're talking about Apache 1.x or 2.x without multithreading, or some older versions of IIS, no.
Apache 1.x forks child processes. AFAIK, 2.x does the same. On *nix, any process can be scheduled on any CPU in the system. So your argument doesn't hold water. *nix has been inherently "multi-threaded" or "multi-tasking" for a very long time, on the order of 30 years.
If you're arguing about Apache Win32 and how Windows handles processes and threads, then it doesn't matter, as Windows sucks as a dedicated server platform, web or otherwise, and isn't worth discussion.
Should have put this in my first post. The thermal conductivity of copper is 401 W/mK, concrete is 0.42 W/mK. This gives us a ratio of approx 950:1, close to my guess of 1000:1. This according to: http://www.engineeringtoolbox.com/thermal-conductivity-d_429.html
Concrete is 950 times less efficient than copper. I don't think the author has enough pipe. He may end up adding a small rad and fan in the basement on either the input or output pipe, doesn't really matter as it's a closed loop. Of course, this is all dependent on his PC, and he didn't state if it's a monster gaming system with SLI and a huge overclock or just an average browser/email box.
Talk about a heat sink. I'm a little surprised that this sort of technique is not more widely adopted at places like data centers
Because concrete has horrible thermal conductivity. Its thermal resistance is orders or magnitude greater than a copper CPU heatsink (without a fan)--we're talking around 1000 fold. Given this gigantic thermal resistance, six meters (20 feet) will likely not be enough to dissipate the heat load from the CPU and/or GPU. The starting, or at rest, temperature of the concrete (61F) is irrelevant. What's critical is how quickly it can absorb the heat out of the piping. If it can't absorb it fast enough, the water temperature will continue to rise, eventually thermaling the CPU/GPU.
Lets say the heat the PC is dumping into the water loop is 100W. If the concrete can only absorb 0.001W/cm^2/sec (possibly a high estimate) from the copper pipe, he'll need 10,000 cm2 of pipe surface area, or 10.76 ft^2, if my math is correct. If he's using standard 1/2" copper plumbing pipe, his surface area for 6 meters is approximately 2.62 ft^2, less than 1/3rd of the approx surface area needed. The thermal resistance of the concrete is dependent on the mix the contractor pours. I have no idea what the exact thermal resistance of the author's concrete will be, but the number with be astronomically high compared to numbers we're all familiar with for copper and aluminum.
Something the author didn't think about, apparently, is that concrete cures via a chemical process that actually _generates_ heat until it is cured. So he'd best plan on not firing up his water loop until after the foundation is fully cured. If you recall, they cast such pipes into the Hoover dam (or was it the Glen Canyon damn?) to do exactly the opposite of what the author is attempting. They ran a water loop through a multi thousand ton freon chiller setup to suck the heat out and cut the curing time of the concrete from 100 years (natural cure time) to just a year or so. Granted that was a few million yards^3 of concrete, but you get the point.
I hope it works out. My gut instinct and rough calculations say 20 ft of pipe won't be enough, unless maybe it's 3/4" or 1" pipe in the slab. I hope we see a future article with results.
I'd like you to hand in your geek card, real geeks read the comic books, THEN saw the movies.
I was aiming for the low hanging fruit Zelucifer. If s/he hasn't seen the movies, s/he def hasn't heard of (or read) the comics. However, given pop culture media and advertising being _everywhere_, s/he must have at least heard of the movies.
You've made an incorrect assumption about me because you didn't comprehend my initial post.
X-gene? I think you mean the Y-chromosome.
Kratisto can't be a geek. "Mutant X gene", Kratisto, as in the X-Men movies. Please drop off your geek badge on the way out the door.
Despite all my efforts, I just can't seem to get a Funny mod. Stolen from me again. I guess now I know what it feels like to be Pauly Shore.
I'd live for that day.
As long as the dead body is inside your door transom, it's self defense. Just make sure you replace the calculator in his/her hand with an ice pick, knife, of pistol. The police will more readily accept your self defense claim with a weapon in the dead hand of the 4'9" slight build brunette female on the floor. Trust me on this.
And when they keep calling you at your job, and insist on speaking to your boss, and call your family, and eventually physically show up at your door, what then?
Two words: Murder
But machetes don't run out of ammunition.
Damn, and I was just going to say minigun for use against hordes...
I was purely aiming for +2 or +3 Funny. Looks like I missed the mark.
They may not use the Linux kernel, but OS X includes a metric shedload...
I believe the quantity you were searching for is 'metric fuckload'. One metric fuckload is greater than a metric shedload by a factor of 10. And there are at least two metric fuckloads of GNU code running around all over the *nix landscape.
Parents, see what happens when you send your kids to public schools? They can't even learn fucking quantities... ;)
B&W use it for their tweeters in their high end speakers, as an example.
They're merely coating an aluminum or titanium diaphragm with diamond dust though, not making diamonds. IIRC they use a super high tech press and mold. The press generates something insane like 100,000 PSI, which fuses the diamond shards to each other and the aluminum diaphragm, creating a smooth solid diamond surface on the concave side of the diaphragm. The result is an extremely light yet rigid diaphragm that doesn't color the high frequencies as much as lesser technologies.
Article contains the terms "ATM Machine" and "PIN Number". Read at your own risk.
People - and by this I mean people on Slashdot, I've not seen anyone complain about it elsewhere - always complain about that. But what's the alternative?
It could be referred as "Personal Identification Number" which is just overly long and besides, everybody just knows it as PIN. They could just say "it would scan their card information and record the PINs they entered" but I don't think it is very good. I know the capitalization makes the necessary difference between "pins" and "PINs" here but honestly, that version still looks a bit out of place to me.
One could say "PIN code". It is the version usually used here in Finland ("PIN-koodi") but the difference to PIN number gets very small.
PIN isn't just an acronym for Personal Identification Number. It is, in itself, a name for a short, usually 4 to 8 digits long digit based password. I could bet a lot of money that most of people don't convert the acronym to words when they read text.
Besides, the ATM machine is used what, once? Most of the time it uses just ATM.
With the massive amount of acronyms we have, especially short ones, a lot of them have multiple meanings. While it is relatively easy to understand these ones in this context, I fully support people adding an additional word to tell which meaning of some acronym is meant in a given situation. At least once in an article. There has been too many times I've seen some acronym, tried to google it, found a dozen different meanings and have had no idea of which it refers to.
This is covered in 9th grade English composition class here in the U.S.--or at least it was when I was in 9th grade.
At the beginning of one's literary piece/article, one introduces the full title of a term to be abbreviated by acronym later in the piece. In other words, you spell it out the first time you make reference to the term at the top of the work/article, then you use the acronym in the rest of the work. Anyone sitting in front of a PC should know this. It's a farking world wide literary composition standard.
According to their marketing dept., rather.
According to TFA, he's not comparing apples to apples. He doesn't state that he verified each burn after the burn, 6-7-8-9 years ago. Thus, there is no way to know whether the bad discs were bad 9 years ago (i.e. a bad burn), or went bad over time. Given the physics and materials involved, I'd say all his errors today were immediately noticeable after the initial burn. He just didn't bother to check. His methodology isn't even close to scientific, anecdotal at beast. This kid is merely yelling "fire" in the CD-R theater.
to a cat, human owner is a mother. they exhibit all behavior they do to their mothers to their human owners.
My cat has never latched onto my nipple to nurse (thank God), so I'd say your argument needs a little refinement.
Ahh, ok. Now we're on the same page. TFA is about vehicle fuel, and thus I thought you were stating the 'locomotive' fuel of the airship was hydrogen. You're absolutely correct regarding hydrogen being what 'fueled' the catastrophic fire.
If the fuel in the Hindenburg had been uncontained gasoline rather than hydrogen...
Hydrogen wasn't used as fuel in the Hindenberg. It was the lifting gas: http://en.wikipedia.org/wiki/Lifting_gas
http://en.wikipedia.org/wiki/Airship
I think you've been caught in that situation where you're talking about something you really don't know enough about. Some folks call that bullshitting. Thus, I call bullshit.
Google simply want to wring every last flop/dollar (TCO) out of their systems...
I think you meant MIP, not FLOP. Search engines don't require floating points calculations. It's all integer baby.
Sounds like a misconfiguration. Some dnsbls explicitly state NOT to use them for URL scoring or blocking due to this very reason. AFAIK, there are only a handful of dnsbls worldwide that can actually be used this way properly.
Whether it's a fair world or not, blacklisting entire blocks and not just the bot-infected or spammer hosts does more harm than good, especially when you're talking about blocking entire netblocks that cross multiple businesses full of non-spammer customers.
Unfortunately one of the roles a dnsbl plays in society is carrying a big stick. Listing large'ish blocks makes legit customers complain to their host. If the host doesn't eject the bad apples from their network, they may lose all those legit customers. Yes, unfortunately, the innocent bleed in the process as well.
Back in the day, SPEWS was effective at this, and network operators were forced to be vigilant in keeping spammers off their nets. Nearly everyone hated SPEWS, but, it was effective. Aggressive dnsbls don't fight as much to keep spam out of inboxen as they fight to get spammers booted from their host.
Stopping spammers is not a technical problem, and cannot be accomplished by technical means. It is a social problem, and business problem, and can only be solved by stopping the spammers from being able to send their spew--preempted at the source. The only way to do this is to keep them off the networks. Legislation is also useless. dnsbls are a critical component of this effort. However, not all dnsbls are equal, and their criteria for listing can be very different.
When you need to scream at someone, scream first at the receiver who rejected your mail. HE rejected it, not the dnsbl. The dnsbl is a tool only and provides data. The receiving mail admin makes the decision to reject an email based on that data. If that admin tells you "buzz off spammer", then scream to your provider to get the real spammers off the network. In other words, YOU have to spend some time doing research in the fight, not just screaming that it's someone else's problem that your email was blocked. The fight against spam includes us ALL. It's going to waste some of your time. Blame this on the spammers, not the dnsbls.
My brother's domain got randomly blacklisted, as did another business venture I'm involved with.
dnsbls are databases of IP addresses. They do not contain domains (example.tld). Thus, your brother's domain did not get black listed. And, contrary to your ignorance, there is no 'randomness' to a block list entry. Each IP address that goes into the database goes there for a very specific reason. Usually the reason is spam sent to a spam trap address from said IP address.
Care to be specific around your actual circumstances.
I get the odd call from a Postini client who's been on mxtoolbox.com, crying, "why am I blacklisted? zomg!". SORBS == idiots.
Really? Idiots? I've never received a legit email from Postini, but I've received tons of spam from them. I'd say in this case SORBS is on the money.
lusers and spammers seem to congregate at the same watering holes, intermixed with one another: Postini, Yahoo, Hotmail, AOL, etc. When the only weapon you have is a shotgun, you're bound to hit some lusers even though you're aiming at the spammers.
filter *URL's pointing to a PBL'd IP that are embedded in a message*!!!
My university does that, too. I run a student organization site that has a university subdomain, but is hosted on a shared host. The host inexplicably got listed in the CBL several times, and that screwed up email for the organization staff, and mailing lists for hundreds of students for days at a time.
I didn't realize anyone else used this brilliant filtering scheme.
So, you admit it's a shared server. It gets listed. So, instead of actually finding out what caused the problem, likely a bot infection, you blame the entity which alerted you to said problem.
When you take your car to the shop, and the mechanic finds something wrong with the car that you didn't already know about, do you then blame the mechanic for the problem?
The primary function of dnsbls is to help prevent spam from reaching inboxen. A very important secondary function of dnsbls is to alert good citizens to the fact that their host is spewing, whatever the reason. It's up the the owner of the listed host to dig in and find out what's wrong. Too many people are too fucking lazy to bother these days. So, they just blame the dnsbl as the source of the problem, when in fact, the dnsbl is merely delivering the message that something is wrong.