Using software RAID5/6 in such a scenario is even worse than RAID1 in a 3-way mirror config, due to the RAID write hole it introduces, you lose terribly, if, say the power fails. Using PCIe controllers with hardware RAID would massively increase the cost of the box.
Hardware RAID controllers have exactly the same issue. They won't save you from the problem you're describing.
That's why it's called silent data corruption RAID5/6 implementations do not read bits from all 3/4 stripes when a read is requested, that would be slow, even slower than it is due to the XOR computation against 2 stripes; they distribute their reads.
Actually it wouldn't make any meaningful difference at all. Parity calculations on any remotely modern CPU are so fast (gigabytes/sec) that dealing with the paltry few 10s - or maybe 100s in rare circumstances - of megabytes per second of data coming off the disks is insignificant.
To be able to verify in software, you'd have to break the RAID abstraction and somehow read bits from each underlying disk, perform both XORs, compare them, alert if they are different. And somehow pull those bits from the other box that doesn't detect an error.
Or you could just keep checksums at the file (rather than stripe) level on several machines and compare them (and play the probabilities game that fewer are corrupted). Which I *hope* is what Backblaze are doing, because otherwise their system is doomed to catastrophic and dramatic failure.
Socialist policies -- although all warm and fuzzy and humanitarian in theory -- fail to address one major flaw: if people don't have to work to take care of themselves because the government will do it for them, then who is going to pay the bills when everyone is staying at home expecting a government handout?
If you think they all suck, how is it FireFox's killer feature then?
Because I'm not talking about the built-on one, I'm talking about the Session Saver Extension (thought it appears to have been renamed Session Manager).
It does. Both the Firefox instances I use on my work PC (I have a FireFox Portable install for "non-work browsing") are lucky to make it through 24 hours without keeling over at least once.
However, with that said, one of them is 11 windows, about 175 tabs and has a memory footprint of about 1.1GB, and the other is 5 windows, about 80 tabs and a memory footprint of about 500MB.
Fortunately, I have the SessionSaver extension (Firefox's killer feature) installed, so a browser crash is mostly an annoyance.
Sure, many of the basic principles don't change, but even so... how do you get your head around writing multi-threaded code for a modern game, when the last thing you learned was Hello World in Fortran?
Completing the last 3 years of your degree would probably be a good place to start.
Nobody is suggesting that two people that are dwarves cannot procreate. Just brothers and sisters. You keep wanting to expand the restriction, when nobody is agreeing to it in the discussion. I am not even suggesting that two people that have Down Syndrome cannot procreate either.
I think you guys need to go back to your basic assumption that "sex" and "procreation" are synonyms.
Whereas everyone else sees contribution of code as a nice bit of corporate philanthropy, Greg KH sees something completely different. He sees it as corporations dumping their code on the community so they can off-load its support.
Which seems rather odd, given that one of the most common arguments put forth from the OSS community is that by open-sourcing their code, commercial vendors will save money because the community will keep them maintained.
When you can define it in an objective fashion, you're on the way to being able to quantify whether or not a given entity can experience it. Before then, you're just waving your arms about.
"Oh but you must believe in God and christianity and stuff because you believe souls exist". Pah...
However, you certainly must believe in the supernatural, which is basically just another word for "God and christianity and stuff".
The creator's family is actually a pretty decent reason. (And, I suspect, the reason that the law was put in place.)
Only if you think companies should be legally required to continue to pay their employee's salaries after they die.
Consider two people, one who creates tangible goods, such as gas stations, and one who creates intellectual property, such as novels. If both of them died today, what would they leave behind in their estate? Of course, both would leave behind any financial earnings from their business as well as things that they had purchased. But what about each person's 'legacy.'
This is simply a matter of timing. If the gas station owner dies while he's still owing a large proportion of the initial expenses endured acquiring it, then his family will get squat. On the other hand, if the author dies after releasing his first blockbuster (but while finishing the second), then his family are going to get a lot of money.
To discourage big media companies from ordering hits ON, rather than FOR their stars.
To what benefit is killing the copyright owner ? That just immediately puts their material into the public domain, effectively destroying any possibility of making money from it.
To say nothing of the laughable idea that someone would choose to engage in the much more serious crime of premeditated murder rather than the typically civil offense of copyright infringement.
My i7 920 system, overclocked to 3.2GHz, draws 95W at idle (monitor excluded). This is based on the APC utility that monitors my UPS unit into which my computer is plugged.
That seems extremely high for an idle system. You should check it with something like a Kill-a-watt.
Are you sure you don't have a few instances of Prime95 or something running in the background ?:)
A modest system can be built in the under 50W range, where gamer systems don't have the 1000W PSU for nothing.Mostly they do. You need an insanely powerful system to come even close to needing a 1kW PSU. A quite capable gaming system will comfortably function with a PSU of half that capacity.
Slashdot Mirror
The RAID5/6 write hole is closed on hardware controllers. Usually by a battery-backed cache module.
No, the probability is somewhat reduced. The problem itself is fundamental and inescapable.
Using software RAID5/6 in such a scenario is even worse than RAID1 in a 3-way mirror config, due to the RAID write hole it introduces, you lose terribly, if, say the power fails. Using PCIe controllers with hardware RAID would massively increase the cost of the box.
Hardware RAID controllers have exactly the same issue. They won't save you from the problem you're describing.
That's why it's called silent data corruption RAID5/6 implementations do not read bits from all 3/4 stripes when a read is requested, that would be slow, even slower than it is due to the XOR computation against 2 stripes; they distribute their reads.
Actually it wouldn't make any meaningful difference at all. Parity calculations on any remotely modern CPU are so fast (gigabytes/sec) that dealing with the paltry few 10s - or maybe 100s in rare circumstances - of megabytes per second of data coming off the disks is insignificant.
To be able to verify in software, you'd have to break the RAID abstraction and somehow read bits from each underlying disk, perform both XORs, compare them, alert if they are different. And somehow pull those bits from the other box that doesn't detect an error.
Or you could just keep checksums at the file (rather than stripe) level on several machines and compare them (and play the probabilities game that fewer are corrupted). Which I *hope* is what Backblaze are doing, because otherwise their system is doomed to catastrophic and dramatic failure.
Try one of these babies on for size. 67TB for about $8,000.
Although, if you want a solution that's fast and reliable, you probably shouldn't.
Socialist policies -- although all warm and fuzzy and humanitarian in theory -- fail to address one major flaw: if people don't have to work to take care of themselves because the government will do it for them, then who is going to pay the bills when everyone is staying at home expecting a government handout?
This is called Begging the Question.
It's called apt. It's already widely deployed in Debian and Ubuntu, and has been for a long time. The problem is solved.
What proportion of third party vendors distribute their software using apt ?
Most economic theories have a hole you could drive a truck through. They assume people act rationally.
More importantly, they assume people are well informed.
Given mortality rates being the highest for drivers 16-24, what would be a better alternative?
Actual driver training that might reduce the accident rate rather than just attempts to apportion blame better ?
If you think they all suck, how is it FireFox's killer feature then?
Because I'm not talking about the built-on one, I'm talking about the Session Saver Extension (thought it appears to have been renamed Session Manager).
Newsflash, it's 2009, every modern browser (that I've used anyway) has this; [...]
And, like the one that's built-in to Firefox, they all suck at it.
Firefox crashes? This is news to me.
It does. Both the Firefox instances I use on my work PC (I have a FireFox Portable install for "non-work browsing") are lucky to make it through 24 hours without keeling over at least once.
However, with that said, one of them is 11 windows, about 175 tabs and has a memory footprint of about 1.1GB, and the other is 5 windows, about 80 tabs and a memory footprint of about 500MB.
Fortunately, I have the SessionSaver extension (Firefox's killer feature) installed, so a browser crash is mostly an annoyance.
A sample size of one is enough to disprove inherent behavior or numerous counter-examples are enough?
Do the words "false dichotomy" mean anything to you ?
Sure, many of the basic principles don't change, but even so... how do you get your head around writing multi-threaded code for a modern game, when the last thing you learned was Hello World in Fortran?
Completing the last 3 years of your degree would probably be a good place to start.
Nobody is suggesting that two people that are dwarves cannot procreate. Just brothers and sisters. You keep wanting to expand the restriction, when nobody is agreeing to it in the discussion. I am not even suggesting that two people that have Down Syndrome cannot procreate either.
I think you guys need to go back to your basic assumption that "sex" and "procreation" are synonyms.
Whereas everyone else sees contribution of code as a nice bit of corporate philanthropy, Greg KH sees something completely different. He sees it as corporations dumping their code on the community so they can off-load its support.
Which seems rather odd, given that one of the most common arguments put forth from the OSS community is that by open-sourcing their code, commercial vendors will save money because the community will keep them maintained.
They didn't donate this code out of any altruism, only pure self-interest.
Everyone using the GPL is doing so out of self-interest. If it was altruism, they'd be released their code into the public domain.
Yeah right. I think your inductive reasoning is more than a little bit rusty if you think a sample size of one says anything about inherent behavior.
If the behaviour described here was inherent, then the numerous counter-examples in this thread could not happen.
Wow, "it works for me!" is now comprehensive proof, eh?
It's all you need to disprove it as inherent behaviour.
I think you are the one 'spitballing' here - inconceivable that some configurations will have problems while others don't...
It certainly seems to be to the poster who started this thread.
I have no problems administering Unix/Linux boxes 200 miles away, but -- MS PR aside -- I'd want to keep a Windows box 'close to home'.
An thanks to ILOM, DRAC, et al, I have no problem looking after Windows machines on the other side of the planet. What's your point ?
Look up 'qualia' on Wikipedia...
When you can define it in an objective fashion, you're on the way to being able to quantify whether or not a given entity can experience it. Before then, you're just waving your arms about.
"Oh but you must believe in God and christianity and stuff because you believe souls exist". Pah...
However, you certainly must believe in the supernatural, which is basically just another word for "God and christianity and stuff".
I have wood working machines from the early 1900's that are more durable, accurate, and mammoth than the cheap plastic shit you buy today.
And were probably priced (relatively speaking) to match, with fewer capabilities.
If this means we actually start over-engineering and building things to -last- again, I'm all for it.
You can buy stuff that lasts today, but it'll cost you (and probably won't have as many features).
The creator's family is actually a pretty decent reason. (And, I suspect, the reason that the law was put in place.)
Only if you think companies should be legally required to continue to pay their employee's salaries after they die.
Consider two people, one who creates tangible goods, such as gas stations, and one who creates intellectual property, such as novels. If both of them died today, what would they leave behind in their estate? Of course, both would leave behind any financial earnings from their business as well as things that they had purchased. But what about each person's 'legacy.'
This is simply a matter of timing. If the gas station owner dies while he's still owing a large proportion of the initial expenses endured acquiring it, then his family will get squat. On the other hand, if the author dies after releasing his first blockbuster (but while finishing the second), then his family are going to get a lot of money.
To discourage big media companies from ordering hits ON, rather than FOR their stars.
To what benefit is killing the copyright owner ? That just immediately puts their material into the public domain, effectively destroying any possibility of making money from it.
To say nothing of the laughable idea that someone would choose to engage in the much more serious crime of premeditated murder rather than the typically civil offense of copyright infringement.
My i7 920 system, overclocked to 3.2GHz, draws 95W at idle (monitor excluded). This is based on the APC utility that monitors my UPS unit into which my computer is plugged.
That seems extremely high for an idle system. You should check it with something like a Kill-a-watt.
Are you sure you don't have a few instances of Prime95 or something running in the background ? :)
A modest system can be built in the under 50W range, where gamer systems don't have the 1000W PSU for nothing.Mostly they do. You need an insanely powerful system to come even close to needing a 1kW PSU. A quite capable gaming system will comfortably function with a PSU of half that capacity.
To be fair, taking off your shoes isn't exactly dependent on some patented expensive technology, either.
It does, however, increase the time it takes a person to go through by roughly an order of magnitude.