In recent months when people talk about a "war on religion" they're more likely to be talking about the HHS mandate that any Calholic who owns a business must violate the teachings of their religion if the want to be allowed to hire employees.
Well, I'd say religious killings are still illegal, but so what. There's plenty of common-good laws that are at odds with a particular religious view. It's neither new nor should it be considered a "war" on religion. The thinking you allude to is same nuttery as if a true-to-form Muslim said that laws that make murder a crime were somehow a "war on Islam", the latter promising that a religious killing is a noble thing to do (in a nutshell). Sorry, most popular religions have aspects of them that are against common good of mankind. It's a fact of life. Get over it, or do something to change it -- meaning, reform the religion, unless you advocate placing your own worldview over the welfare of others...
I'd have hoped I was obvious enough. In order to prevent from I*R voltage drops, you put a box in a box. With insulator between them. Faraday cage inside of a Faraday cage. When the exterior one gets hit by lightning, there'll be perhaps nasty I*R drops, imagine a continuous metal wall suddenly developing, say 10V/cm potential difference along the direction of lightning current's average flow. How is your capacitor rant relevant I don't see, neither do I see why you thought I implied open boxes...
Personal anecdote re.2, and I hope I'm not mixing it up too much. I eventually understood it in linear algebra. The integration happens to work like a dot product of two vectors, if you consider linear vector space of functions. If you think of abstract vector spaces, it becomes (eventually) quite blatant that it must be so. To me the big point was wrapping my head around the fact that "vectors of numbers" are merely representations of abstract vectors in a certain base, and that the abstract vectors are just that. It so happens that functions form vector spaces too. The set of exponentials of a certain form happen to form a base of the vector space of functions. The frequency domain happens to be equivalent to a certain orthonormal basis. Other interesting domains (like quefrency) have their own bases as well.
It being an antenna is irrelevant. The effects we're talking about are close to DC. It's not an antenna but a large single-turn transformer winding. So, what it is, is a bunch of perfectly kosher low impedance loops. Every pair of phase wires (or phase+neutral) makes such a beautiful loop, stretching over miles of countryside. Give the loop enough loop area, let the *large* magnetic field change, but slowly, and voila -- the loop is a big transformer, and slowly changing (50Hz) currents get induced in it. The rest of the power grid is not designed to cope with such "almost" DC currents and promptly indicates the same by, in minor cases, tripping a lot of breakers as the control circuits detect unwanted DC -- thus blackouts.
In more major cases, the breakers will trip but won't interrupt the current because their interrupting capacity at DC is *way* smaller than at AC. A breaker with 100kA interrupting capacity at 50Hz may have only 10kA interrupting capacity at DC (or even less), as an example. Some breakers have no interrupting capacity at DC to speak of -- even a small current, like 1A on a 100kA breaker, may well generate a self-sustaining arc. So if you put more than 100kA DC through it, it will trip, but it won't do anything. Heck, at say 50kA DC it won't trip, but the transformers may well have their cores getting driven into saturation and getting much hotter than the RMS current would indicate. When the control system or operators decide they have to trip the breakers to protect the transformers, the breakers will open, but the current won't stop flowing.
Every electrical engineer probably knows this perfectly well. Everyone who isn't one -- well, good luck with that 100kA 30kVDC current source. You can't exactly buy it in Walmart. The explanation is perfectly useless to anyone who doesn't have specialist knowledge. It's like detailed instructions for a thermonuclear weapon. Good luck making one if you don't already know how (or most of it, anyway).
Agreed. I don't see the value of Facebook on student-accessible computers. As for the teachers, they should have access to everything. Anything else would be stupid. It's an education of learning, you can't a priori decide that some things have no educational value. Besides, why on earth ban Facebook use during teacher's off time. I mean, give me a break, you already provide teachers with a lounge, perhaps a cafeteria, etc. Barring recreational internet access on school grounds makes no sense to me at all.
The nun is partly right, partly wrong. Yes, restrictions will exacerbate the problem. No restrictions, though, won't make the problem magically go away either. I mean, there *is* a problem to begin with -- that they'll run into porn, or whatever else passes for inappropriate content. Porn-wise, I think that kids who are raised in a home where nudity is no big deal will react appropriately: shrug it off, saying "so what, haven't you seen a naked guy/girl?!". Sex isn't exactly a visually engaging thing if you don't pay much attention to nudity to begin with. Up to a certain age, at least, I'd think. In homes where privates were verboten to see except by yourself in the mirror -- oh well, those will be the problem kids. There's no way to ensure, much less be sure of, "all our girls [being] good".
Our typical power distribution infrastructure is designed to cope with 50/60Hz loads. Any overload at 0.1Hz or less will likely destroy it. At such frequencies the transformers turn into resistors (core saturation), the breakers turn into slightly dissipative shunts, and a lot of fires and explosions ensue. If you want to blow up a power distribution transformer, temporarily connect a DC current source providing 3-5x rated winding RMS current to the middle phase. The compliance of the current source probably needs to be "only" up to 4-8x the RMS voltage rating, that should be enough to keep the breakers arcing over. In a minute or so the transformer is gone (or going and self sustaining), and you can connect to the next one to keep at it.
It would only be bad for the power distribution infrastructure because said infrastructure isn't designed to deal with "almost DC" induced voltages. All the breakers likely won't interrupt large DC currents at all -- there will be a sustained arc that won't extinguish that will keep the current flowing. The transformers will get destroyed when their cores saturate and due to DC currents flowing through the phase conductors. Upon saturation, the transformer is no more inductive, it becomes a resistor that converts all the incoming energy to heat. Realistically, that means an oil explosion in short order in oil cooled types. Air cooled ones will catch fire when the insulation is hot enough to combust (the copper will heat it up, so it's not like there'd be much in the way of heatsinking going on).
Besides, grounding etc. is only effective at fairly low frequencies (100s of kHz to perhaps single MHz). If the transient is fast enough, then the grounding lead is open circuit anyway.
This is bullshit. You don't need to ground the Faraday cage. It doesn't matter what is the potential between the cage and the ground/earth. It's a self-contained system, the inside is shielded from the outside. Lightning protection and grounding is only relevant when the system has connections to the outside. If all you want is an isolated Faraday cage, you don't need anything special. Make it from thick enough steel such that it will decently carry lightning currents with "acceptable" resistive voltage drop along the internal surface, and that's about it. For all I care it can be two layers with an insulator between them, so that even large I*R voltage drops on the outside won't propagate to the inner layer (other than by electromagnetic induction -- in that case increase suitably the insulating gap).
Hmm, that's useful to know. Maybe I was unnecessarily blaming st driver code where it might be a bad interaction between that and the scsi host driver. Alas, none of my servers use scsi host adapters anymore.
We're talking about a very specific thing -- namely the debts owed to government. That's not something that there's a big point in hiding. If their reports say those debts are repaid, and they weren't, it'd be too low hanging of a fruit for a lot of regulators not to act swiftly. Enron, Worldcom and France Telecom were a very different kind of animal; I have heard some first hand accounts from people who were the to clean up the mess. I'm not familiar with what AIG and Tyco was all about.
That's a crazy argument. With a classroom sample of 30 kids or so, you can't pretend that you're separating anything. For all I know you've got a bunch of doofuses, with crappy socioeconomic background to boot, with firm beliefs that the adults in the school are teh devil that has to "dealt with". For all we know they're fucking the teacher's dog every time he grades them less than A, and their future goal is to be the best crack distributor for the next 5 blocks up and down the main street. Everyone got their own goals, you know.
Quality of the student is separate from quality of instruction, yet I see no reason why a poorly instructed student should get good grades without putting in extra effort (if it's possible at all). Sure it's not their *fault* they get bad grades due to bad instruction, but the fact is, their knowledge is subpar. Grades reflect knowledge, not effort, goodwill nor any other politically correct notion du jour. You go to a bad school and suffer for it -- too bad, the fact is you are a bad student without any fault of yours. Low grades don't imply fault with student, perhaps the idiots who think that should get off their high horse. Correlation not being causation and all that jazz.
Perhaps the issue is that it'd be nice to have two components of a grade: grades that score knowledge on a uniform, repeatable scale, and grades that reflect the work put in. I don't see, for example, that a student who brilliant in maths and everything there comes easy to him shoudn't get an A for knowledge and D for effort. As to what use such grades are, it's a different story. Perhaps this allowed him to redirect his efforts towards English Lit where he got a solid B for knowledge putting A+ of effort.
It's not a Dell problem per se. The linux st driver needs a rewrite that's long overdue. The Dell should of course know better than to sell stuff that's obviously rather poorly tested. Same issues reproduce on same Adaptec cards in any other machine, so it's hardly Dell that'd be to blame. I haven't tried RHEL6 since I have given up by then. For a while the workaround was to link the RHEL5 server to a RHEL4 server with a dedicated gigabit link and have the tape attached to a RHEL4 server. After a while of this I said "fuck this, it's supposed to be simple, with all the hoops there's no point". And that was it.
I was running tape "solutions" sold by brand name vendors, all bought with the server, on a single PO. All the parts were supposed to have been certified to work together. I had more success with old aluminum-backed-tape integral capstan tape cartridges and SCSI-1 tape drives pulled from junked servers. Those seemed to work without a hitch. Modern LTO tape systems are seemingly a pile of junk in comparison -- I don't care how well they might work if the planets are aligned just so, all I had to work with was a supported, off-the-shelf "solution". I survived three generations of LTO with constant headaches until I called it quits.
Holy batman, how the fuck does tape labelling and drive "management" help (or not) with anything? What the heck do you do to those tape drives, open them up, load a test tape, hook up to an oscilloscope and tweak trimmers per a service manual?! Tape drives are pretty much a finite life, disposable item. You read so many tapes between cleanings, and then after so many cleanings they are junk and you replace them. They keep internal maintenance counters, for crying out loud -- at least they did a couple years ago when I got tired of that shit. Managing them is trivial, at least if they are supported on the OS you're using. I used Dell-branded drives and the firmware of everything from the drive to the motherboard was updated using yum. No magic there. If that doesn't work, there's nothing else to do but toss the whole non-solution out of the window. It's supposed to just work, and it mostly did except that all parts of the system always seems to be on the verge of not working. I never trusted tape because of that. I'd trust a dropped hard drive more than that.
It's a whole different story when it comes to tape contents. If you forget how you formatted whatever was written to the tape, you can mostly kiss it good bye. A tape is logically a couple partitions, that are logically a bunch of blocks. It's like the mainframe days all over again. I'll take a hard drive with a filesystem and descriptive filenames any day over a tape. Mismatched tape hardware versions -- fuck, man, either it's compliant to a certain standard or it's not. If an up-to-date LTO-5 drive with up-to-date firmware doesn't read LTO-5 tapes without handwaving, then in my book it's all a junk of a standard. Again -- I would have hoped this kind of handholding was gone with the heydays of big iron. How the heck a hard drive, with much higher complexity, can be a plug-and-play item, where with tapes you have to muck around with every little detail or else.
Never mind that at least in RHEL5/CentOS5, for a good period of time the scsi tape driver was unusable on popular Dell server hardware of the time. I consider the scsi tape driver in the Linux kernel to be still broken, for all practical purposes. It sucks.
For an HDD "robot", it makes no sense to handle the drives. It's cheap enough to route the SATA signals around a bit with all drives plugged in and have a single PCI-X interface chip per a set of drives. The drives can be (should be!) spun down or even powered off when not in use.
The modern drives will deal with a stretched tape I'm sure without much issue. In a linear-scan tape, reading a fraction (say 1/8th or 1/16th) of the tape's capacity will go end-to-end anyway, as data is recorded on a set of tracks and when you hit end of the tape, the head indexes to a new set of tracks, the tape reverses, and keeps on going.
All I remember is that my tape adventure ended with an LTO-3 drive that lasted about a year, doing daily backups, on a 7 tape daily rotation. I had two LTO/DLT generations before that, and a prior helical scan system with smaller tapes as well. The Linux kernel drivers for tape devices never worked all that well -- getting good throughput required tweaking and I had to add a large buffer between gzip and tape in spite of a fast machine. The particular drive in question worked fine on RHEL4 but would not work on a Dell server under RHEL5 where the throughput went to hell -- it was 5-8x slower. If you plan on using any tape drives, make sure that you test the exact combination of server, interface, interconnect, drive, tape, Linux distribution and backup software/scripts you're going to run it as. Anything less may lull you into a false sense of accomplishment that will be blown away in short order once it fails in production.
In times of linux kernel 2.0 I had some junk machines that had ISA-based Adaptec SCSI cards and worked quite well with old 80 megabyte aluminum-plate-backed tape cartridges. It was slow, but it still kept the tape streaming at full speed. It was foolproof, and redirecting tar -z output to/dev/st0 was all that was needed to get it to work, after setting desired block size. Even on a 486-class machine. It seems that it was downhill ever since in terms of ability for the systems to maintain streaming and retain data. I can still read those 80 megabyte tapes with zero errors. All of a dozen of them or so. All of the newer tapes would develop errors under normal use, sometimes after a couple of uses!
Slashdot Mirror
In recent months when people talk about a "war on religion" they're more likely to be talking about the HHS mandate that any Calholic who owns a business must violate the teachings of their religion if the want to be allowed to hire employees.
Well, I'd say religious killings are still illegal, but so what. There's plenty of common-good laws that are at odds with a particular religious view. It's neither new nor should it be considered a "war" on religion. The thinking you allude to is same nuttery as if a true-to-form Muslim said that laws that make murder a crime were somehow a "war on Islam", the latter promising that a religious killing is a noble thing to do (in a nutshell). Sorry, most popular religions have aspects of them that are against common good of mankind. It's a fact of life. Get over it, or do something to change it -- meaning, reform the religion, unless you advocate placing your own worldview over the welfare of others...
I'd have hoped I was obvious enough. In order to prevent from I*R voltage drops, you put a box in a box. With insulator between them. Faraday cage inside of a Faraday cage. When the exterior one gets hit by lightning, there'll be perhaps nasty I*R drops, imagine a continuous metal wall suddenly developing, say 10V/cm potential difference along the direction of lightning current's average flow. How is your capacitor rant relevant I don't see, neither do I see why you thought I implied open boxes...
Pray tell explain the mechanism of grounding helping with anything.
Personal anecdote re.2, and I hope I'm not mixing it up too much. I eventually understood it in linear algebra. The integration happens to work like a dot product of two vectors, if you consider linear vector space of functions. If you think of abstract vector spaces, it becomes (eventually) quite blatant that it must be so. To me the big point was wrapping my head around the fact that "vectors of numbers" are merely representations of abstract vectors in a certain base, and that the abstract vectors are just that. It so happens that functions form vector spaces too. The set of exponentials of a certain form happen to form a base of the vector space of functions. The frequency domain happens to be equivalent to a certain orthonormal basis. Other interesting domains (like quefrency) have their own bases as well.
You're an idiot.
Yeah, but a bunch of modbus stuff going over a radio link isn't exactly hard to spoof, ya know.
It being an antenna is irrelevant. The effects we're talking about are close to DC. It's not an antenna but a large single-turn transformer winding. So, what it is, is a bunch of perfectly kosher low impedance loops. Every pair of phase wires (or phase+neutral) makes such a beautiful loop, stretching over miles of countryside. Give the loop enough loop area, let the *large* magnetic field change, but slowly, and voila -- the loop is a big transformer, and slowly changing (50Hz) currents get induced in it. The rest of the power grid is not designed to cope with such "almost" DC currents and promptly indicates the same by, in minor cases, tripping a lot of breakers as the control circuits detect unwanted DC -- thus blackouts.
In more major cases, the breakers will trip but won't interrupt the current because their interrupting capacity at DC is *way* smaller than at AC. A breaker with 100kA interrupting capacity at 50Hz may have only 10kA interrupting capacity at DC (or even less), as an example. Some breakers have no interrupting capacity at DC to speak of -- even a small current, like 1A on a 100kA breaker, may well generate a self-sustaining arc. So if you put more than 100kA DC through it, it will trip, but it won't do anything. Heck, at say 50kA DC it won't trip, but the transformers may well have their cores getting driven into saturation and getting much hotter than the RMS current would indicate. When the control system or operators decide they have to trip the breakers to protect the transformers, the breakers will open, but the current won't stop flowing.
Every electrical engineer probably knows this perfectly well. Everyone who isn't one -- well, good luck with that 100kA 30kVDC current source. You can't exactly buy it in Walmart. The explanation is perfectly useless to anyone who doesn't have specialist knowledge. It's like detailed instructions for a thermonuclear weapon. Good luck making one if you don't already know how (or most of it, anyway).
Agreed. I don't see the value of Facebook on student-accessible computers. As for the teachers, they should have access to everything. Anything else would be stupid. It's an education of learning, you can't a priori decide that some things have no educational value. Besides, why on earth ban Facebook use during teacher's off time. I mean, give me a break, you already provide teachers with a lounge, perhaps a cafeteria, etc. Barring recreational internet access on school grounds makes no sense to me at all.
The nun is partly right, partly wrong. Yes, restrictions will exacerbate the problem. No restrictions, though, won't make the problem magically go away either. I mean, there *is* a problem to begin with -- that they'll run into porn, or whatever else passes for inappropriate content. Porn-wise, I think that kids who are raised in a home where nudity is no big deal will react appropriately: shrug it off, saying "so what, haven't you seen a naked guy/girl?!". Sex isn't exactly a visually engaging thing if you don't pay much attention to nudity to begin with. Up to a certain age, at least, I'd think. In homes where privates were verboten to see except by yourself in the mirror -- oh well, those will be the problem kids. There's no way to ensure, much less be sure of, "all our girls [being] good".
Our typical power distribution infrastructure is designed to cope with 50/60Hz loads. Any overload at 0.1Hz or less will likely destroy it. At such frequencies the transformers turn into resistors (core saturation), the breakers turn into slightly dissipative shunts, and a lot of fires and explosions ensue. If you want to blow up a power distribution transformer, temporarily connect a DC current source providing 3-5x rated winding RMS current to the middle phase. The compliance of the current source probably needs to be "only" up to 4-8x the RMS voltage rating, that should be enough to keep the breakers arcing over. In a minute or so the transformer is gone (or going and self sustaining), and you can connect to the next one to keep at it.
It would only be bad for the power distribution infrastructure because said infrastructure isn't designed to deal with "almost DC" induced voltages. All the breakers likely won't interrupt large DC currents at all -- there will be a sustained arc that won't extinguish that will keep the current flowing. The transformers will get destroyed when their cores saturate and due to DC currents flowing through the phase conductors. Upon saturation, the transformer is no more inductive, it becomes a resistor that converts all the incoming energy to heat. Realistically, that means an oil explosion in short order in oil cooled types. Air cooled ones will catch fire when the insulation is hot enough to combust (the copper will heat it up, so it's not like there'd be much in the way of heatsinking going on).
Besides, grounding etc. is only effective at fairly low frequencies (100s of kHz to perhaps single MHz). If the transient is fast enough, then the grounding lead is open circuit anyway.
This is bullshit. You don't need to ground the Faraday cage. It doesn't matter what is the potential between the cage and the ground/earth. It's a self-contained system, the inside is shielded from the outside. Lightning protection and grounding is only relevant when the system has connections to the outside. If all you want is an isolated Faraday cage, you don't need anything special. Make it from thick enough steel such that it will decently carry lightning currents with "acceptable" resistive voltage drop along the internal surface, and that's about it. For all I care it can be two layers with an insulator between them, so that even large I*R voltage drops on the outside won't propagate to the inner layer (other than by electromagnetic induction -- in that case increase suitably the insulating gap).
Hmm, that's useful to know. Maybe I was unnecessarily blaming st driver code where it might be a bad interaction between that and the scsi host driver. Alas, none of my servers use scsi host adapters anymore.
We're talking about a very specific thing -- namely the debts owed to government. That's not something that there's a big point in hiding. If their reports say those debts are repaid, and they weren't, it'd be too low hanging of a fruit for a lot of regulators not to act swiftly. Enron, Worldcom and France Telecom were a very different kind of animal; I have heard some first hand accounts from people who were the to clean up the mess. I'm not familiar with what AIG and Tyco was all about.
That's a crazy argument. With a classroom sample of 30 kids or so, you can't pretend that you're separating anything. For all I know you've got a bunch of doofuses, with crappy socioeconomic background to boot, with firm beliefs that the adults in the school are teh devil that has to "dealt with". For all we know they're fucking the teacher's dog every time he grades them less than A, and their future goal is to be the best crack distributor for the next 5 blocks up and down the main street. Everyone got their own goals, you know.
Quality of the student is separate from quality of instruction, yet I see no reason why a poorly instructed student should get good grades without putting in extra effort (if it's possible at all). Sure it's not their *fault* they get bad grades due to bad instruction, but the fact is, their knowledge is subpar. Grades reflect knowledge, not effort, goodwill nor any other politically correct notion du jour. You go to a bad school and suffer for it -- too bad, the fact is you are a bad student without any fault of yours. Low grades don't imply fault with student, perhaps the idiots who think that should get off their high horse. Correlation not being causation and all that jazz.
Perhaps the issue is that it'd be nice to have two components of a grade: grades that score knowledge on a uniform, repeatable scale, and grades that reflect the work put in. I don't see, for example, that a student who brilliant in maths and everything there comes easy to him shoudn't get an A for knowledge and D for effort. As to what use such grades are, it's a different story. Perhaps this allowed him to redirect his efforts towards English Lit where he got a solid B for knowledge putting A+ of effort.
Huh? They are a publicly traded company. They have financial reporting, and a lot of it, too.
It's not a Dell problem per se. The linux st driver needs a rewrite that's long overdue. The Dell should of course know better than to sell stuff that's obviously rather poorly tested. Same issues reproduce on same Adaptec cards in any other machine, so it's hardly Dell that'd be to blame. I haven't tried RHEL6 since I have given up by then. For a while the workaround was to link the RHEL5 server to a RHEL4 server with a dedicated gigabit link and have the tape attached to a RHEL4 server. After a while of this I said "fuck this, it's supposed to be simple, with all the hoops there's no point". And that was it.
I was running tape "solutions" sold by brand name vendors, all bought with the server, on a single PO. All the parts were supposed to have been certified to work together. I had more success with old aluminum-backed-tape integral capstan tape cartridges and SCSI-1 tape drives pulled from junked servers. Those seemed to work without a hitch. Modern LTO tape systems are seemingly a pile of junk in comparison -- I don't care how well they might work if the planets are aligned just so, all I had to work with was a supported, off-the-shelf "solution". I survived three generations of LTO with constant headaches until I called it quits.
Holy batman, how the fuck does tape labelling and drive "management" help (or not) with anything? What the heck do you do to those tape drives, open them up, load a test tape, hook up to an oscilloscope and tweak trimmers per a service manual?! Tape drives are pretty much a finite life, disposable item. You read so many tapes between cleanings, and then after so many cleanings they are junk and you replace them. They keep internal maintenance counters, for crying out loud -- at least they did a couple years ago when I got tired of that shit. Managing them is trivial, at least if they are supported on the OS you're using. I used Dell-branded drives and the firmware of everything from the drive to the motherboard was updated using yum. No magic there. If that doesn't work, there's nothing else to do but toss the whole non-solution out of the window. It's supposed to just work, and it mostly did except that all parts of the system always seems to be on the verge of not working. I never trusted tape because of that. I'd trust a dropped hard drive more than that.
It's a whole different story when it comes to tape contents. If you forget how you formatted whatever was written to the tape, you can mostly kiss it good bye. A tape is logically a couple partitions, that are logically a bunch of blocks. It's like the mainframe days all over again. I'll take a hard drive with a filesystem and descriptive filenames any day over a tape. Mismatched tape hardware versions -- fuck, man, either it's compliant to a certain standard or it's not. If an up-to-date LTO-5 drive with up-to-date firmware doesn't read LTO-5 tapes without handwaving, then in my book it's all a junk of a standard. Again -- I would have hoped this kind of handholding was gone with the heydays of big iron. How the heck a hard drive, with much higher complexity, can be a plug-and-play item, where with tapes you have to muck around with every little detail or else.
Never mind that at least in RHEL5/CentOS5, for a good period of time the scsi tape driver was unusable on popular Dell server hardware of the time. I consider the scsi tape driver in the Linux kernel to be still broken, for all practical purposes. It sucks.
For an HDD "robot", it makes no sense to handle the drives. It's cheap enough to route the SATA signals around a bit with all drives plugged in and have a single PCI-X interface chip per a set of drives. The drives can be (should be!) spun down or even powered off when not in use.
The modern drives will deal with a stretched tape I'm sure without much issue. In a linear-scan tape, reading a fraction (say 1/8th or 1/16th) of the tape's capacity will go end-to-end anyway, as data is recorded on a set of tracks and when you hit end of the tape, the head indexes to a new set of tracks, the tape reverses, and keeps on going.
All I remember is that my tape adventure ended with an LTO-3 drive that lasted about a year, doing daily backups, on a 7 tape daily rotation. I had two LTO/DLT generations before that, and a prior helical scan system with smaller tapes as well. The Linux kernel drivers for tape devices never worked all that well -- getting good throughput required tweaking and I had to add a large buffer between gzip and tape in spite of a fast machine. The particular drive in question worked fine on RHEL4 but would not work on a Dell server under RHEL5 where the throughput went to hell -- it was 5-8x slower. If you plan on using any tape drives, make sure that you test the exact combination of server, interface, interconnect, drive, tape, Linux distribution and backup software/scripts you're going to run it as. Anything less may lull you into a false sense of accomplishment that will be blown away in short order once it fails in production.
In times of linux kernel 2.0 I had some junk machines that had ISA-based Adaptec SCSI cards and worked quite well with old 80 megabyte aluminum-plate-backed tape cartridges. It was slow, but it still kept the tape streaming at full speed. It was foolproof, and redirecting tar -z output to /dev/st0 was all that was needed to get it to work, after setting desired block size. Even on a 486-class machine. It seems that it was downhill ever since in terms of ability for the systems to maintain streaming and retain data. I can still read those 80 megabyte tapes with zero errors. All of a dozen of them or so. All of the newer tapes would develop errors under normal use, sometimes after a couple of uses!
One can do Earth-field NMR with "RF" at 2kHz (audio frequencies!), see for example Magritek's Terranova-MRI.