It seems to me, from reading the original article, that what the poster really wants is side-mounted access ports (sound, (tv) video, USB, etc) a side/top
drive bay is just an easy way to get that. (there's not much reason to have
a DISK bay side mounted, much less top mounted)
A guy that's trying to solicit a young boy over the 'net, turns
out to be the coach of a young boys team. I'd say that it's a
relatively high probability that he's attempting to do things
with the boys on his team -- and it's probably more likely that
he'll succeed there, than on the net. Parents tend to tell their
kids to trust the coach. They're much less likely to let
them trust some random identity on the 'net.
I'd consider those kids to be in a high risk situation, and
would stongly encourage that someone be sent to debrief
them and make sure that nothing happened (or, if something did
happen, find out what happened and deal with it).
The most surprising and welcome news of the afternoon was that, despite the alarmist title, there was not a panicked call for additional legislation.. . . . in fact, that Rep. Bass seemed a little bored and annoyed. He had to remind everyone twice that he and his colleague were lawmakers:". . ..If you could draft the bill, what would it say?" (emphasis mine)
Clearly the intent of the hearings was to generate interest in a bill.
Granted, most of the witnesses and even most of the parents at the hearings were intelligent enough to realize that extra legislation was not the best answer to this problem, but if Bass has his way, he will be producing some legislation that he can bandy about -- Legislation that is pretty likely to be of borderline constitutionality.. possibly outlawing the use of IRC, or something stupid like that.
It's not that he's intent on crushing our civil rights -- He's more likely interested in safeguarding his legislative seat.. Our rights would simply be collateral damage in the resulting stampede.
Someone needs to get to him and convince him that the hearings are a valuable end in, and of, themselves. Given that education has been most often touted as the true solution, he should be trumpeting himself as the person who made that need so plainly known. Perhaps he could even get a patronage job for a friend's kid making sure that the information gets out to as many parents as possible.
Surely this sort of thing happens all the time in chatboxes.
The press are able to get to the unusual cases. They tend to broadcast the unusual cases because they are unusual. You don't hear a big media to-do about kids who drown in their bathtubs, or fall down the stairs and kill themselves, or die in car accidents because that does happen more often. Often enough that it's not longer big news.
The fact that there's a big media blitz about it should be an indicator that it probably is an unusual occurence.
For me, the real scary thing is that this apparent child-predator was the trainer of a boy's football team. That may be something that they stepped over because it was just too damned common.
I'd be calling the TV station who broke the story and asking them if they arranged cautionary counselling for 'his boys' on the football team. Chances are that they didn't, and that's where I see the real threat.
From his homapage:
There are more than 100,000 customers including many well known ones
like IBM, Xerox [who made RealNames partner of the year last year],
EBay, Mattel - who have Keywords on every Barbie Box, and many more.
....
What can you do? Probably nothing.
I think that there is something that people can do.
Create their own name tool.
It seems to me that there are enough 'big movers' in this process that
a consortium to re-install a naming process into IE is possible.
Not only that, but it could be done in an 'open' manner such that the
same naming mechanism could be used for IE, Netscap, Mozilla and any
other browser that was interested in doing so.
Yes, this might require that realnames restart it's process, to a
certain extent, but they will have to do this anyways if the company
is to thrive. Microsoft is *NOT* necessary to this. They were the
best way to get the process kick-started. Now that people know what
realnames is capable of, it's possible to now take this to the next
level -- but without any fealty payments to Microsoft.
This could be the death of realnames, or it could be a new beginning.
If realnames really wants to take on this task, one of the first things to do would
probably be to create an add-on/plugin, and put some add hooks into the
links created by real-names such that people know where to find the
new extension. Then people at various large sites would need to
put links allowing people to find the addin as well.
Time is short, but the opportunity is as large is the problem.
Tesla wasn't a genius. More like a crackpot with all his deathrays and "unlimited energy" ideas.
Yeah... That, and the completely insane idea that AC current was more efficient than DC current.
Tesla and Edison probably split because of their differening ideas on the efficacy of AC current vs. DC. Edison firmly believed that DC was the way to go and did Everything in his power to destroy Tesla and his AC ideas. Like Bill Gates, however, he was able to recognize when an idea was stronger than his marketing clout and turn to an 'embrace and extend' approach.
Tesla died a (relative) pauper because he was willing to give away his AC patents to General Electric rather than see Edison win with his DC ideas. Without Tesla, we would not have things like transformers, AC motors and long-distance power transmission. Edison, on the other hand, had money and marketing on his side.
Perhaps Microsoft is learning something from their interaction with NW school districts. Even better yet: It appears that school districts have learned something too.
Do you really want to risk having your name on the comet that wipes out New York state? It's enough to give the paranoid pause for thought...
Re:Caffeine toxicity - material safety data
on
The Plague of Frogs
·
· Score: 2
As has been said many times, LD50 is the dose at which 1/2 the population would die. THis means that, for someone like me (who avoids coffee/caffine), sucking back 5 cups over a few minuts might very well kill me. Other people who regularly drink 8-10 cups a day have obviously developed a tolerance, and might even suffer mild withdrawal effects from reducing themselves to 5 cups/day.
Actually, I'd be inclined to say eight. There are 11 untimely deaths, but only 8 are truely mysterious.
The guy who got offed by his daughters and her satanist friends is reasonably well explained, and "normal" in that -- If you're going to get offed, chances are it's going to be by family or 'friends'. You even have living, breathing conspirators who, if they were part of a larger comspiracy, would probably be happy to talk about it. The only thing really unusual about it is the satanist connection.
Similarly, the Pizza delivery murder-suicide is similarly well explained and fits the family/friend statistical norm. In this case, however, everybody who was directly involved is dead. Unless someone can point to some evidence that the suicide was 'forced', I'm going to chaulk this one up to a love triangle, or something and mark it 'untimely but explained'.
3 untimely death in a short time period is probably not that far off for a group of this size. As for the other 8 deaths, I'm betting that they take this group far outside the statistical norm.
(I don't mind copying this, because I feel it needs to be moderated up (from 0) and, given that an AC posted it, I'm don't risk stealing karma from anybody) Anonymous Coward wrote:
Note: I'm a radioastronomer involved in instrumentation.
Half a femtowatt is still quite a lot of power in
a 1kHz bandwidth (you want to have some redundancy
in the 770 bits, so let us round it up to 1kHz).
Let us start from the noise power of a resistor
at room temperature in 1kHz bandwidth, it is about -144dBm or 4e-17 W, or 40 attowatts. Now building narrowband amplifiers whose effective noise temperature is much lower than 300K is a no brainer (for people who know how to build amplifiers around discrete pseudomorphic
high electron mobility transistors). I've not checked recently what the room temperature performance of the most recent transistors is
but I'd believe that a noise figure of 0.5 dB
(equivalent to the noise of a resistor at 40 Kelvin) is routinely guaranteed at 12 GHz on
commercial transistors, like Fujitsu's FHX13. As I said better models might be available now.
In a 1KHz bandwidth the equivalent input noise power of an amplfier like his ould be around
5 attowatts.
Now if you push me, I'd tell you that by carefully selecting the transistors and operating them at cryogenic temperature, the amplifier noise will be even lower, perhaps by a factor 5 or so (over 3 guaranteed).
At this point the most difficult problem
even with a high quality antenna is probably to get rid of human interference and noise is probably dominated by the scattering of the antenna beam. Note that I do millimeter radioastronomy so I don't know well the detailed technical problems of centimeter guys, but while we don't (yet) have problems of interference with
other human activities, it is a severe problem at longer wavelengths.
In any case, you have a reasonable signal to noise ratio, even if the antennas are smaller than what you think (DSN antennas are 60 or 64 meters, I can't remember the exact value).
Ok, that's all, but remember that we radioastronomers use the Jansky as flux measurement unit. A Jansky is 1E-26 W/Hz/m,
so it would be about 0.01 attowatt on a DSN antenna in 1 KHz BW. And 1 Jy is a strong source,
even in the millimeter range, we are rather chasing the milliJansky (on wide bandwidth however).
I've been trying for the last hour to connect to my favorite mirror for a couple of RPMs that I need, and it keeps coming back full... but that's never happened in recent memory. My favorite server's almost always completely unloaded (that's why it's my favorite!). Then I think: "This feels like a slashdot. I wonder if the 7.3 release has just been announced?".
Sure enough.......
Maybe I'll get my RPMs tomorrow (or tonight at 4am).
"What this decision means," she added, "is that one side of the debate gets full free speech protection, but a corporation trying to defend itself is subject to strict liability."
At the risk of being taken to court.... bullshit!
What this decisionn means is that if I knowingly lie about Nike's policies and procedures, they can take me to court (and often will).
On the other hand, if Nike knowingly lies about Nike's policies and procedures, I can take them to court.
Pretty much an even playing field, in my mind (except for the fact that they have 12,000 times the legal budget that I do.)
This is really very much like a libel or slander case. The company is speaking about a specific party (themselves) with speech that is likely to cause injury to someone (the public).Although the parties affected by the 'libel' are unusual, I think that it's quite reasonable to apply the legal principles around libel to a case such as this.
If Nike had won this case, it would meant that corporations would be able to lie through their teeth with impunity.
The court made a distinction:
When talking about general public policy issues, Nike has the right to say whatever they want with full first amendment protections.
When speaking about NIKE policy issues, they have to be truthful.
This is not an onerous or unreasonable demand: Nike has a unique ability to know definitively what their own policy and aproaches are, and -- for the most part -- the public generally has to accept what they say as the truth because it's usually rather difficult to do the work it takes to independently investigate the truth of such statements.
If what a company is doing is nasty, underhanded and disgusting, they may also be uniquely motivated to lie about what they're doing.
This also doesn't specially bind the company in the debate:
If I knowingly lie about what the company is doing, then I'm subject to a Libel suit (even if I'm not lieing, I could be subject to a SLAPP suit).
Similarly, if the company knowingly lies about what they're doing, then they should be subject to a libel suit (or something equivalent to it). This rulling essentially says that they are.
If you break the GPL, then your rights revert to what you would have with any other piece of software under regular copyright (i.e. right to use, but not to copy).
For 'regular people', a short explanation of the GPL should be enough.
If your business license is going to survive erring on the side of give-away with GPL code, then you should have nothing to worry aobut. If you're going to try and play footsie with the GPL and mix it with proprietary code, it would be stupid to spend $1M+ on the venture based on your reading of a 3 line summary of a 3 page document. Grab a good lawyer and go over the thing line by line. If If still scares you, after that, then you probably should be scared.
The GPL does not impact users of the software, only distributors. This isn't strictly true. Software development organizations....
Software development organizations are distributers, not just users -- so they are one of the exceptions to which the GPL applies.
If You give me a copy (modified or not), then you're a distributer and subject to the GPL. As a simple user, I don't have to worry about the GPL until somebody asks me for a copy -- at which point I have to either
give them a copy of everything I got -- including a pointer to where the source is available (if I didn't get that), or:
Just give them the source code, or:
Give them some excuse, like "I'm too busy today".
Oh yeah: they also need to get a copy of the license, somewhere in there, if I choose 1 or 2.
If you follow the right of First Sale, then you don't have to agree to the GPL.
This would, however, require that, when you sell your copy, you give the recipient ALL VERSIONS of your copy, and destroy any backups you may have.. If you kept/gave a copy then you've made a copy and that action would be subject to regular copyright restrictions (unless you've agreed to the GPL).
That would apply in cases like a contractor doing a work-for-hire on a piece of GPL work, and leaving the result with the employer. -- but you'd still have to give them your source code, because if you gave them a modified object, it would be a copy, and subject to copyright rules (i.e. you can't). On the other hand: if you gave them an unmodified object/source, what in the world did they pay you for?
Even systems that are multi-user are rarely 100% loaded at any given time. That's why we can get away with making multi-user systems. A system with continuous 100% utilization is probably ill-provisioned (unless the 100% is because of low-priority idle-tasks like seti-at-home). I've seen systems with 300 users still sitting below 50% usage for reasonable periods of time.
Although single user systems can have multiple processes at any given time, many/most single user apps usually have one process (max) accessing the disk at a time. A UNIX box being used as a dedicated network log would be such an application.
Granted -- a one-line summary doesn't include the source code, but it does indicate the kind of testing methadology they're likely to be using.
(btw: I consider my own box single user (me) even though it has 100% utilization -- I run Seti-at-home. The disk utilization is still single-user in nature (seti rarely accesses the disk)
Nevertheless, it does seem to be the ugly truth, at least for straightforward read/write tests in a single-user environment.
You essentially have one process doing sequential writes to disk.
As almost everybody who has commented intelligently on the difference between SCSI and IDE seems to agree, this is the kind of situation that's going to favor IDE.
I'd go as far as to say that it's probably close to IDE's dream application.
No big surprise here, for me.
ON the other hand, 25,000 users trying to POP their mail will probably beat those same IDE drives into submission faster than you can say "thrash the system".
It's not just buffering. The SCSI protocol allows you to queue a whackload of write requests and then go get more data while the disk is processing your queue and then queue a bit more.
With IDE, you have to have the data waiting and ready to go before the hardware buffer on the drive runs out. If you have any sort of latency problems you're more likely to die with IDE than with SCSI -- it's because of what the protocol allows you to do with your data/commands.
Those sorts of problems aren't as critical in normal disk I/O situations, but they will show up as hickups in the data flow if you're pushing the disks/controller trying to get all of your data to/from the disk in time -- especially if you're seeking all over the disk.
As if the tens of thousands of times this has been hashed out weren't enough already...
In something like logging network traffic, SCSI would give pretty
much zero advantage for the extra cost. I'm only vaguely surprised
that IDE is faster than SCSI for raw speed.
My analogy:
IDE is like a Yamaha motercycle -- motor, wheels, seat. little else.
SCSI is like a BMW sedan -- more weight but more volume too.
If all you want to do is get one or two people (400LB biker and his
girlfriend) from point 'A' to point 'B' as fast as possible, the
Yamaha/IDE may be your best bet.
On the other hand, when you've got a spouse, 2 screaming Kids, a dog
and a cooler fool of picnic supplies, I'd still suggest the BMW/SCSI.
Network logging is a 400lb biker type of situation. 100,000 users trying to access their e-mail would be a screaming kids application.
MTBF would, as you said, be average time before data loss to the entire array -- (and would, nominally, presume that you get the replacement drive running within the granularity of the measurement (1 hour, in this case)). That implies a hot-spare pool.
On the other hand, if you presume 24 hours to get the replacement disk in and populated, then you'd be looking at 5,000*5,000 days == 25million days , or "ONLY" ABOUT 68K YEARS.
Like said -- still more likely that the silicon is going to wear out with those sorts of times.
However, if the two disks which die hold the same data, that data is lost....
That presumes that your raid 1 mirrors are only two plexes each... There's nothing wrong with doing (say) a 7 way mirror. It'll be hell on writes without hardware support, but really nice for a read-mostly database with lots of concurrent access -- and you can lose up to 6 disks, guaranteed, before you risk real data loss (as opposed to performance changes).
I vaguely remember doing a 4 way mirror one day, using the veritas volume manager, when I had a 'spare' 10 disk box sitting about to be put into service. It was mostly just proof-of-concept. I didn't do any real benchmarking because I didn't have any real applications that might have needed that kind of that kind of data flow/reliability. (or, for that matter, enough spare time)
Slashdot Mirror
Is this correct?
I'd consider those kids to be in a high risk situation, and would stongly encourage that someone be sent to debrief them and make sure that nothing happened (or, if something did happen, find out what happened and deal with it).
in fact, that Rep. Bass seemed a little bored and annoyed. He had to remind everyone twice that he and his colleague were lawmakers:". . .
Clearly the intent of the hearings was to generate interest in a bill.
Granted, most of the witnesses and even most of the parents at the hearings were intelligent enough to realize that extra legislation was not the best answer to this problem, but if Bass has his way, he will be producing some legislation that he can bandy about -- Legislation that is pretty likely to be of borderline constitutionality.. possibly outlawing the use of IRC, or something stupid like that.
It's not that he's intent on crushing our civil rights -- He's more likely interested in safeguarding his legislative seat.. Our rights would simply be collateral damage in the resulting stampede.
Someone needs to get to him and convince him that the hearings are a valuable end in, and of, themselves. Given that education has been most often touted as the true solution, he should be trumpeting himself as the person who made that need so plainly known. Perhaps he could even get a patronage job for a friend's kid making sure that the information gets out to as many parents as possible.
The press are able to get to the unusual cases. They tend to broadcast the unusual cases because they are unusual. You don't hear a big media to-do about kids who drown in their bathtubs, or fall down the stairs and kill themselves, or die in car accidents because that does happen more often. Often enough that it's not longer big news.
The fact that there's a big media blitz about it should be an indicator that it probably is an unusual occurence.
For me, the real scary thing is that this apparent child-predator was the trainer of a boy's football team. That may be something that they stepped over because it was just too damned common.
I'd be calling the TV station who broke the story and asking them if they arranged cautionary counselling for 'his boys' on the football team. Chances are that they didn't, and that's where I see the real threat.
There are more than 100,000 customers including many well known ones like IBM, Xerox [who made RealNames partner of the year last year], EBay, Mattel - who have Keywords on every Barbie Box, and many more.
....
What can you do? Probably nothing.
I think that there is something that people can do.
Create their own name tool.
It seems to me that there are enough 'big movers' in this process that a consortium to re-install a naming process into IE is possible. Not only that, but it could be done in an 'open' manner such that the same naming mechanism could be used for IE, Netscap, Mozilla and any other browser that was interested in doing so.
Yes, this might require that realnames restart it's process, to a certain extent, but they will have to do this anyways if the company is to thrive. Microsoft is *NOT* necessary to this. They were the best way to get the process kick-started. Now that people know what realnames is capable of, it's possible to now take this to the next level -- but without any fealty payments to Microsoft.
This could be the death of realnames, or it could be a new beginning.
If realnames really wants to take on this task, one of the first things to do would probably be to create an add-on/plugin, and put some add hooks into the links created by real-names such that people know where to find the new extension. Then people at various large sites would need to put links allowing people to find the addin as well.
Time is short, but the opportunity is as large is the problem.
Yeah... That, and the completely insane idea that AC current was more efficient than DC current.
Tesla and Edison probably split because of their differening ideas on the efficacy of AC current vs. DC. Edison firmly believed that DC was the way to go and did Everything in his power to destroy Tesla and his AC ideas. Like Bill Gates, however, he was able to recognize when an idea was stronger than his marketing clout and turn to an 'embrace and extend' approach.
Tesla died a (relative) pauper because he was willing to give away his AC patents to General Electric rather than see Edison win with his DC ideas. Without Tesla, we would not have things like transformers, AC motors and long-distance power transmission. Edison, on the other hand, had money and marketing on his side.
Perhaps Microsoft is learning something from their interaction with NW school districts. Even better yet: It appears that school districts have learned something too.
Do you really want to risk having your name on the comet that wipes out New York state? It's enough to give the paranoid pause for thought...
As has been said many times, LD50 is the dose at which 1/2 the population would die. THis means that, for someone like me (who avoids coffee/caffine), sucking back 5 cups over a few minuts might very well kill me. Other people who regularly drink 8-10 cups a day have obviously developed a tolerance, and might even suffer mild withdrawal effects from reducing themselves to 5 cups/day.
Actually, I'd be inclined to say eight. There are 11 untimely deaths, but only 8 are truely mysterious.
The guy who got offed by his daughters and her satanist friends is reasonably well explained, and "normal" in that -- If you're going to get offed, chances are it's going to be by family or 'friends'. You even have living, breathing conspirators who, if they were part of a larger comspiracy, would probably be happy to talk about it. The only thing really unusual about it is the satanist connection.
Similarly, the Pizza delivery murder-suicide is similarly well explained and fits the family/friend statistical norm. In this case, however, everybody who was directly involved is dead. Unless someone can point to some evidence that the suicide was 'forced', I'm going to chaulk this one up to a love triangle, or something and mark it 'untimely but explained'.
3 untimely death in a short time period is probably not that far off for a group of this size. As for the other 8 deaths, I'm betting that they take this group far outside the statistical norm.
Any epidemiologists out there?
Anonymous Coward wrote:
Note: I'm a radioastronomer involved in instrumentation.
Half a femtowatt is still quite a lot of power in a 1kHz bandwidth (you want to have some redundancy in the 770 bits, so let us round it up to 1kHz).
Let us start from the noise power of a resistor at room temperature in 1kHz bandwidth, it is about -144dBm or 4e-17 W, or 40 attowatts. Now building narrowband amplifiers whose effective noise temperature is much lower than 300K is a no brainer (for people who know how to build amplifiers around discrete pseudomorphic high electron mobility transistors). I've not checked recently what the room temperature performance of the most recent transistors is but I'd believe that a noise figure of 0.5 dB (equivalent to the noise of a resistor at 40 Kelvin) is routinely guaranteed at 12 GHz on commercial transistors, like Fujitsu's FHX13. As I said better models might be available now. In a 1KHz bandwidth the equivalent input noise power of an amplfier like his ould be around 5 attowatts.
Now if you push me, I'd tell you that by carefully selecting the transistors and operating them at cryogenic temperature, the amplifier noise will be even lower, perhaps by a factor 5 or so (over 3 guaranteed).
At this point the most difficult problem even with a high quality antenna is probably to get rid of human interference and noise is probably dominated by the scattering of the antenna beam. Note that I do millimeter radioastronomy so I don't know well the detailed technical problems of centimeter guys, but while we don't (yet) have problems of interference with other human activities, it is a severe problem at longer wavelengths.
In any case, you have a reasonable signal to noise ratio, even if the antennas are smaller than what you think (DSN antennas are 60 or 64 meters, I can't remember the exact value).
Ok, that's all, but remember that we radioastronomers use the Jansky as flux measurement unit. A Jansky is 1E-26 W/Hz/m, so it would be about 0.01 attowatt on a DSN antenna in 1 KHz BW. And 1 Jy is a strong source, even in the millimeter range, we are rather chasing the milliJansky (on wide bandwidth however).
Sure enough.......
Maybe I'll get my RPMs tomorrow (or tonight at 4am).
I don't have the space in my sig to do that... I barely have space for the sig itself.
At the risk of being taken to court.... bullshit!
What this decisionn means is that if I knowingly lie about Nike's policies and procedures, they can take me to court (and often will).
On the other hand, if Nike knowingly lies about Nike's policies and procedures, I can take them to court.
Pretty much an even playing field, in my mind (except for the fact that they have 12,000 times the legal budget that I do.)
If Nike had won this case, it would meant that corporations would be able to lie through their teeth with impunity.
The court made a distinction:
This is not an onerous or unreasonable demand: Nike has a unique ability to know definitively what their own policy and aproaches are, and -- for the most part -- the public generally has to accept what they say as the truth because it's usually rather difficult to do the work it takes to independently investigate the truth of such statements.If what a company is doing is nasty, underhanded and disgusting, they may also be uniquely motivated to lie about what they're doing.
This also doesn't specially bind the company in the debate:
If I knowingly lie about what the company is doing, then I'm subject to a Libel suit (even if I'm not lieing, I could be subject to a SLAPP suit). Similarly, if the company knowingly lies about what they're doing, then they should be subject to a libel suit (or something equivalent to it). This rulling essentially says that they are.
If you break the GPL, then your rights revert to what you would have with any other piece of software under regular copyright (i.e. right to use, but not to copy).
If your business license is going to survive erring on the side of give-away with GPL code, then you should have nothing to worry aobut. If you're going to try and play footsie with the GPL and mix it with proprietary code, it would be stupid to spend $1M+ on the venture based on your reading of a 3 line summary of a 3 page document. Grab a good lawyer and go over the thing line by line. If If still scares you, after that, then you probably should be scared.
Software development organizations are distributers, not just users -- so they are one of the exceptions to which the GPL applies.
If You give me a copy (modified or not), then you're a distributer and subject to the GPL. As a simple user, I don't have to worry about the GPL until somebody asks me for a copy -- at which point I have to either
- give them a copy of everything I got -- including a pointer to where the source is available (if I didn't get that), or:
- Just give them the source code, or:
- Give them some excuse, like "I'm too busy today".
Oh yeah: they also need to get a copy of the license, somewhere in there, if I choose 1 or 2.This would, however, require that, when you sell your copy, you give the recipient ALL VERSIONS of your copy, and destroy any backups you may have .. If you kept/gave a copy then you've made a copy and that action would be subject to regular copyright restrictions (unless you've agreed to the GPL).
That would apply in cases like a contractor doing a work-for-hire on a piece of GPL work, and leaving the result with the employer. -- but you'd still have to give them your source code, because if you gave them a modified object, it would be a copy, and subject to copyright rules (i.e. you can't). On the other hand: if you gave them an unmodified object/source, what in the world did they pay you for?
Although single user systems can have multiple processes at any given time, many/most single user apps usually have one process (max) accessing the disk at a time. A UNIX box being used as a dedicated network log would be such an application.
Granted -- a one-line summary doesn't include the source code, but it does indicate the kind of testing methadology they're likely to be using.
(btw: I consider my own box single user (me) even though it has 100% utilization -- I run Seti-at-home. The disk utilization is still single-user in nature (seti rarely accesses the disk)
Nevertheless, it does seem to be the ugly truth, at least for straightforward read/write tests in a single-user environment.
You essentially have one process doing sequential writes to disk. As almost everybody who has commented intelligently on the difference between SCSI and IDE seems to agree, this is the kind of situation that's going to favor IDE. I'd go as far as to say that it's probably close to IDE's dream application.
No big surprise here, for me.
ON the other hand, 25,000 users trying to POP their mail will probably beat those same IDE drives into submission faster than you can say "thrash the system".
With IDE, you have to have the data waiting and ready to go before the hardware buffer on the drive runs out. If you have any sort of latency problems you're more likely to die with IDE than with SCSI -- it's because of what the protocol allows you to do with your data/commands.
Those sorts of problems aren't as critical in normal disk I/O situations, but they will show up as hickups in the data flow if you're pushing the disks/controller trying to get all of your data to/from the disk in time -- especially if you're seeking all over the disk.
In something like logging network traffic, SCSI would give pretty much zero advantage for the extra cost. I'm only vaguely surprised that IDE is faster than SCSI for raw speed.
My analogy:
IDE is like a Yamaha motercycle -- motor, wheels, seat. little else.
SCSI is like a BMW sedan -- more weight but more volume too.
If all you want to do is get one or two people (400LB biker and his girlfriend) from point 'A' to point 'B' as fast as possible, the Yamaha/IDE may be your best bet.
On the other hand, when you've got a spouse, 2 screaming Kids, a dog and a cooler fool of picnic supplies, I'd still suggest the BMW/SCSI.
Network logging is a 400lb biker type of situation. 100,000 users trying to access their e-mail would be a screaming kids application.
On the other hand, if you presume 24 hours to get the replacement disk in and populated, then you'd be looking at 5,000*5,000 days == 25million days , or "ONLY" ABOUT 68K YEARS.
Like said -- still more likely that the silicon is going to wear out with those sorts of times.
That presumes that your raid 1 mirrors are only two plexes each... There's nothing wrong with doing (say) a 7 way mirror. It'll be hell on writes without hardware support, but really nice for a read-mostly database with lots of concurrent access -- and you can lose up to 6 disks, guaranteed, before you risk real data loss (as opposed to performance changes).
I vaguely remember doing a 4 way mirror one day, using the veritas volume manager, when I had a 'spare' 10 disk box sitting about to be put into service. It was mostly just proof-of-concept. I didn't do any real benchmarking because I didn't have any real applications that might have needed that kind of that kind of data flow/reliability. (or, for that matter, enough spare time)