> 10 PCs that can run, say, Office will be cheaper than one big machine than > can run 10 copies of Office (plus virtualization overhead, of course)
Are you including about hardware, environment, administration, maintenance...? Probably not.
> Further, you need to engineer a lot more reliability into those resources, > because if they go away *everyone* is down, rather than one user.
Except that companies care about man-hours, and all things being equal, it doesn't matter in your example. If you have one 10-user machine that has 1 hour of downtime per year, and 10 1-user machines that have 1 hour of downtime per year, you have, in both cases, 10 man-hours of downtime.
The practical reality is that it's cheaper/more efficient to engineer reliability at a central point, versus at each node. Furthermore, the amount of downtime is a function of two things - failure rate, and time needed to repair a failure. If a system is in a server room, and not in 10 different locations, how do you think that effects the time needed for repair?
Another practical reality is that when you centralize, you need less machine per person, because 10 users are not using 100% of 1 CPU/RAM/disk etc. all the time. You get economy by putting all the CPU, RAM and disk in one place. 2-4 CPUs versus 10 CPUs, in your example.
> The problem, basically, is that IT administrators suck. Address that issue > (with better tools, more admins, better training, whatever)
Hey, thanks:) I personally am comfortable (and trained and equipped, by the way) to work in both types of environment. However, for a business trying to make money, a correctly-engineered thin-client architecture has clear advantages over a thick client setup. The only "problem" is deciding what actually makes sense for your environment.
> A full-powered workstation is *cheaper* than a thin client.
This is usually false, both in terms of hardware cost, lifetime expectancy, power consumption, and deployment cost, yadda yadda. Any way you slice it, a workstation is not cheaper than any but the most unfairly-priced and poorly-designed thin client.
> It's stupid to waste all this computing power, only to channel more and more > money into more and more powerfull app servers.
A bunch of single processer machines, each with its own board, memory, IO, fans, footprint and power supply (w/ AC-DC transformer) is neccessarily more "wasteful" in terms of resources than a WTS running on an SMP machine. That's basic physics. When the cost between one and the other becomes insignificant, then you start to have a point; or if you're rich enough, maybe it doesn't matter. Nowadays, though, it usually does.
> We've got incredibly cheap computing power that would have been unimaginable > even 20 years ago, lets not waste it all - design ways to leverage to power of > workstations while alleviating the administrative overhead.
That's exactly what VM clusters and terminal servers do. For workstations, the best you can do is: imaging, or scripted installs with SMS/Netinstall; either case requires a server infrastructure anyway. So you're back to having thick clients AND extra machines in the back room (which are idle most of the time, like fat clients).
This is an age-old argument, and there are sometimes cases where thick clients are a must-have (3D or even 2D graphics, or non multi-user aware apps, for example). But most users can go without and suffer no loss in productivity; hell, they can even benefit, because it's easier and cheaper to engineer reliability into the system.
> it generally targets the types of people the anti-SUV types don't like, for lack > of a better way of putting it. "Suburban yuppie capitalist conservatives", if > you will.
If a player don't play, there'd be no haters. Class prejudice doesn't spontaneously spring into existence; and while that's no justification for being disdainful toward yuppies a priori (or iced out suburban black guys in Escalades, for that matter), it's worth noting that one neccessarily cause the other.
You want LUNAR. It's especially cool because it uses an underlay network called selnet (ARP forwarding instead of straight-up IP routing). Also, there are a bunch of normal layer 3 ad-hoc multihop protocols designed especially for highly dynamic/mobile nets that you can install for free (I can verify that they work on 2.4 kernels, anyway):
You might want to check this out... there are practical problems with large diameter networks. Some of the problems lay at the MAC layer (i.e. in the specs), so solving them probably requires radically new thinking.
The problem with the rosy view is that most real study has been done in simulation. There are not a lot of papers detailing real, large scale testbeds (with a literal handful of exceptions).
And the airport is nice, but I wouldn't want to participate as a mobile node with that card without an energy-aware network stack - I gather it is one of the worst in terms of power efficiency.
Adama: OK. Send the engineers down to work on the Vipers. Get that limey fellow working with Dr. Baltar on a plan to destroy that Base Star. And send that one to my quarters.
> Thankyou for reproducing what is the crappiest list ever. >...the continuing decendency of/. into the armpit of mediocrity.
No, Thankyou. Maybe they'll increase the frequesncy of good stories when the frequesncy of good posts increases.
What is an armpit of mediocrity, anyway? Is that like the bad part of mediocrity? Is that still better than being on the shoulder of piss-poor? Gosh, I hope I don't descenden to that level.
All TV shows rely on fan charity via one source or another.
If the shows ratings are a good indicator of how many people are watching the commercials and buying the products adverstised, then the "charity" effort will fail.
However, if the ratings samples are NOT representative for some reason (e.g. bitorrent), AND enough people understand that it may not be a lost cause, and someone organizes it properly, it might work. My guess is that it won't, because grassroots support takes a while to filter up to the broadcast execs -- think Family Guy. By the time they mobilize enough fans (assuming there are enough) and convince the head honkey in charge, the series finale will have been aired.
Agreed, there is a difference in the intended application of crypto and steganography - I was just pointing out that steg is not terribly advanced when applied as an encryption, and hence not the "future of encryption" as proposed by OP.
It's exhausting enough being a pedant without having to deal with others:P
So are statist Democrats... look at Joe Lieberman's voting record re: censorship. Al Gore may have invented the Internet, but he also backed the friggin' V Chip.
| The humor is too "in your face," with no subtleties at all.
Ahh - but were there really no subtelties, or did you simply miss them?
[honk, honk]
C'mon... tit, dick and fart jokes can be fun even if you are smart. If I want high-brow, I won't watch the Simpsons either... I'll read Proust or something.
I'm not neccessarily commenting on it's practical effectiveness but isn't this really the past in terms of encryption? As in, Caesar-cipher era - i.e. the method is the key.
Slashdot Mirror
> 10 PCs that can run, say, Office will be cheaper than one big machine than
:) I personally am comfortable (and trained and equipped, by the way) to work in both types of environment. However, for a business trying to make money, a correctly-engineered thin-client architecture has clear advantages over a thick client setup. The only "problem" is deciding what actually makes sense for your environment.
> can run 10 copies of Office (plus virtualization overhead, of course)
Are you including about hardware, environment, administration, maintenance...? Probably not.
> Further, you need to engineer a lot more reliability into those resources,
> because if they go away *everyone* is down, rather than one user.
Except that companies care about man-hours, and all things being equal, it doesn't matter in your example. If you have one 10-user machine that has 1 hour of downtime per year, and 10 1-user machines that have 1 hour of downtime per year, you have, in both cases, 10 man-hours of downtime.
The practical reality is that it's cheaper/more efficient to engineer reliability at a central point, versus at each node. Furthermore, the amount of downtime is a function of two things - failure rate, and time needed to repair a failure. If a system is in a server room, and not in 10 different locations, how do you think that effects the time needed for repair?
Another practical reality is that when you centralize, you need less machine per person, because 10 users are not using 100% of 1 CPU/RAM/disk etc. all the time. You get economy by putting all the CPU, RAM and disk in one place. 2-4 CPUs versus 10 CPUs, in your example.
> The problem, basically, is that IT administrators suck. Address that issue
> (with better tools, more admins, better training, whatever)
Hey, thanks
> A full-powered workstation is *cheaper* than a thin client.
This is usually false, both in terms of hardware cost, lifetime expectancy, power consumption, and deployment cost, yadda yadda. Any way you slice it, a workstation is not cheaper than any but the most unfairly-priced and poorly-designed thin client.
> It's stupid to waste all this computing power, only to channel more and more
> money into more and more powerfull app servers.
A bunch of single processer machines, each with its own board, memory, IO, fans, footprint and power supply (w/ AC-DC transformer) is neccessarily more "wasteful" in terms of resources than a WTS running on an SMP machine. That's basic physics. When the cost between one and the other becomes insignificant, then you start to have a point; or if you're rich enough, maybe it doesn't matter. Nowadays, though, it usually does.
> We've got incredibly cheap computing power that would have been unimaginable
> even 20 years ago, lets not waste it all - design ways to leverage to power of
> workstations while alleviating the administrative overhead.
That's exactly what VM clusters and terminal servers do. For workstations, the best you can do is: imaging, or scripted installs with SMS/Netinstall; either case requires a server infrastructure anyway. So you're back to having thick clients AND extra machines in the back room (which are idle most of the time, like fat clients).
This is an age-old argument, and there are sometimes cases where thick clients are a must-have (3D or even 2D graphics, or non multi-user aware apps, for example). But most users can go without and suffer no loss in productivity; hell, they can even benefit, because it's easier and cheaper to engineer reliability into the system.
> it generally targets the types of people the anti-SUV types don't like, for lack
> of a better way of putting it. "Suburban yuppie capitalist conservatives", if
> you will.
If a player don't play, there'd be no haters. Class prejudice doesn't spontaneously spring into existence; and while that's no justification for being disdainful toward yuppies a priori (or iced out suburban black guys in Escalades, for that matter), it's worth noting that one neccessarily cause the other.
You want LUNAR. It's especially cool because it uses an underlay network called selnet (ARP forwarding instead of straight-up IP routing). Also, there are a bunch of normal layer 3 ad-hoc multihop protocols designed especially for highly dynamic/mobile nets that you can install for free (I can verify that they work on 2.4 kernels, anyway):
NIST AODV
unik OLSR
US NAVY labs OLSR
CLICK modular router (contains a DSDV and DSR implementation, provides a framwork for rapid prototyping of stack behaviors)
These all might be nice for a smallish office as a way to extend and enhance the probable coverage area of the network without getting more APs.
You might want to check this out... there are practical problems with large diameter networks. Some of the problems lay at the MAC layer (i.e. in the specs), so solving them probably requires radically new thinking.
The problem with the rosy view is that most real study has been done in simulation. There are not a lot of papers detailing real, large scale testbeds (with a literal handful of exceptions).
And the airport is nice, but I wouldn't want to participate as a mobile node with that card without an energy-aware network stack - I gather it is one of the worst in terms of power efficiency.
You know, with an attitude like yours, you really missed your calling as a nerd battle rapper.
Adama: OK. Send the engineers down to work on the Vipers. Get that limey fellow working with Dr. Baltar on a plan to destroy that Base Star. And send that one to my quarters.
Chief: Right away, sir.
Typing something IS an action.
Seriously, I'm not saying you're right or wrong in your overall assessment, but using words has effect, like any other action.
daveschroeder's Latest 24 of 709 Comments (516195)
:P
Apparently they're a breeze to administer!
Hear, hear re: the vibrator. It's probably the most blatant oversight in the list.
> Thankyou for reproducing what is the crappiest list ever. ...the continuing decendency of /. into the armpit of mediocrity.
>
No, Thankyou. Maybe they'll increase the frequesncy of good stories when the frequesncy of good posts increases.
What is an armpit of mediocrity, anyway? Is that like the bad part of mediocrity? Is that still better than being on the shoulder of piss-poor? Gosh, I hope I don't descenden to that level.
Try rounding and using simple arithmetic on the powers of 10 in order to check your answer.
round 1,500,000 to 2E6
round 3,600 s/h to 4E3
-> 2E6 * 4E3 = 8E9
round c 186,000 mi/s to 2E5
-> 2E5 / 8E9 = 0.25E-4
percentage conversion *100 is *E2
-> 0.25E-4 * E2 = 0.25E-2 = 0.0025%
Relativistic effects are negligible.
How much do people spend on coffee and soda every day?
This argument is, in this case, very silly. We're talking $10-20 per fan. Can you honestly say you've never spent $10-20 of non-philanthropic money?
All TV shows rely on fan charity via one source or another.
If the shows ratings are a good indicator of how many people are watching the commercials and buying the products adverstised, then the "charity" effort will fail.
However, if the ratings samples are NOT representative for some reason (e.g. bitorrent), AND enough people understand that it may not be a lost cause, and someone organizes it properly, it might work. My guess is that it won't, because grassroots support takes a while to filter up to the broadcast execs -- think Family Guy. By the time they mobilize enough fans (assuming there are enough) and convince the head honkey in charge, the series finale will have been aired.
> luxuries beyond beer seem like a major drain on mankind sometimes...
Did you happen to see the beer commercials during the Super Bowl? The 4.4 million dollar/minute beer commercials?
Agreed, there is a difference in the intended application of crypto and steganography - I was just pointing out that steg is not terribly advanced when applied as an encryption, and hence not the "future of encryption" as proposed by OP.
:P
It's exhausting enough being a pedant without having to deal with others
Yeah, and he's also a Democrat last time I checked. That was my point, sunshine.
That's what I was getting at - that steganography is not encryption, at least not in a modern sense.
So are statist Democrats... look at Joe Lieberman's voting record re: censorship. Al Gore may have invented the Internet, but he also backed the friggin' V Chip.
| The humor is too "in your face," with no subtleties at all.
Ahh - but were there really no subtelties, or did you simply miss them?
[honk, honk]
C'mon... tit, dick and fart jokes can be fun even if you are smart. If I want high-brow, I won't watch the Simpsons either... I'll read Proust or something.
I'm not neccessarily commenting on it's practical effectiveness but isn't this really the past in terms of encryption? As in, Caesar-cipher era - i.e. the method is the key.
> Politically correct term? For an exploding star? Stars have feelings?
No, but the geeks that lose sleep over the precise way to describe the events do.
A stellar explosion is a nova, and this doesn't appear to have been a nova event, but something somewhat less cataclysmic.
> $105 billion.
And since there are, like, a billion people in the USA, freedom costs exactly $1.05. QED
There's something ironic about asking someone if you should be skeptical.
> Quantum crypto algorithms which is a subject I know very little about.
:P
Here you go:
msg XOR key