It is not cost-effective immediately, however adding that much insulation on the house has about a 15-year payback on heating an cooling costs so it is cost-effective in the long run. Rockwool makes the house fireproof too.
I work for Akamai; Akamai does offer a general-transport better-than-BGP service called Sure Route IP.
The idea is that we utilize the massive amounts of data about the Internet's health and the insanely scalable alogorithms for matching end-users to the HTTP server that can best serve them (called mapping) to create generalized IP tunnels that send traffic across "routes" that know more about the Internet then BGP does.
Think about it. ..BGP routes based on the least number of hops. ..there are many problems inherant in that. We route based on ping data, bandwidth, cost, reliability, etc, etc, etc.
Don't worry. ..there is new legislation on the president's desk right now that will make it "virtually impossible now to get a nationwide class-action suit off the ground." -- The Economist
> The only reason I use Firefox (and I use it a > lot) is that I can't split proxy servers in > Opera.
Sure you can. ..you just have to write a proxy.pac (pac = proxy auto-config). It is a javascript file that points your browser to the proper proxy server depending on the request.
For reference:
http://wp.netscape.com/eng/mozilla/2.0/relnotes/ de mo/proxy-live.html
No, you're right. They actually do. At the college I used to go to, for example, they hired and paid a student to be a "Microsoft Evangellist" (they actually titled the position that) whose sole job is to say that Microsoft is great.
Here's a solution to the whole debate -- make the sticky bit have meaning under Linux like it does on other UNIXen -- if the sticky bit is set on the execuatble, do not swap it. If it is not set, the executable is free to be swapped.
This solves the entire debate (for instance, if you don't want the 'interactive' mozilla process swapped, set the sticky bit on the executable).
Dual-head with tv-out so you can have a movie display on the TV while working on your monitor
Very good high TV-out quality (some say it is one of the best out there)
You can pick it up cheap on ebay
Perhaps the best supported graphics hardware under linux
Special support in mplayer for beautiful movie display
Hardware MJPEG compression for capture
Nice break-out-box with RCA in/out as well as svideo in/out
All this in one AGP card
Drawbacks:
Current Linux driver does not do DMA for plain v4l (non-MJPEG) capture -- it does memcpy() which is CPU intensive. I'm looking to add support to the driver at some point when I finish reading the SPECS.
Capture driver does not play well with the framebuffer driver. Again I'm looking to help fix this at some point
The nicest thing about the card is that specs for most of the chips are avaiable if you wish to hack it.
To see more about the linux support:
There are many different implementations of QoS and some solve different problems:
* Some give priority to specific protocols or
ports * Some give priority to smaller packets over
larger * Some give priority based on the ToS bit * Some just drop packets for "high bandwidth"
connections and hope IP drops its window size
There are also about a dozen different ways to do each of the above and each vendor has a different name for it. Thus there is much confusion on this subject and I did not make it clear in my post which I was speaking of.
The idea is to schedule packets based upon origin / destination IP address rather then by a protocol, port or connection. The simplest schemes such as WRR mentioned above, require little CPU because they are simply round-robin. The down-side to this is that they make the TCP window size thrash. More advanced algorithms use a modified token-bucket scheme and grant a specific number of tokens per second to each IP address it sees.
In both cases, because it is done by IP and not by protocol or connection, Joe P Hacker can have 1,000 P2P connections going and if Grandma loads a web page, Joe's connections will not slow her down while still giving Joe all the availible bandwidth he wants. If Grandma and Sue down the street load a web page, again they will load fast (slowing down Joe for a second) and then Joe resumes. If Joe loads a web page while he has 1,000 P2P connections, his own connections will drown him out. If Joe P Hacker is running 1,000 P2P connections while Julie is downloading the latest Linux distro and they are the only two on the wire, they'll get equal bandwidth.
Who cares if the pipe is saturated by P2P people if we can guarantee that everyone else's traffic gets through when they need it?
So this type of a scheme is not a backbone solution, it is a near-leaf solution. ISPs implement this scheme (or at least should) within regions and then balance their regional routers with another scheme more suited to massive bandwidth.
Granted, I have never done this in a NAP scenario, however in a company with ~500 employees after proper tuning I never had to worry about any one person downloading too much crap ever again. But as mentioned before, this would not be done at the NAP anyway -- it should be done more towards the leaves.
Note that I said there are other implementations that scale better in the original post, and that WRR was just an example to illustrate the point.
There are many papers on this subject and many routers implement the protocol -- even to "hundreds of thousands of hosts".
Google for "Weighted Fair Queueing". I don't have the time right now to right a tutorial on the subject. There are papers that answer your other questions.
The solution ISPs can use is weighted fair queueing. This ensures that one person does not destroy the bandwidth of another. It is a hell of a lot better then making users worry about how much they download.
One such implementation is the Weighted Round Robin qdisc in Linux:
http://wipl-wrr.dkik.dk/wrr/
There are other implementations that scale better.
I say this every time someone brings up the "scarce bandwidth" issue, but no one ever listens and ISP continue to use draconian way to solve their bandwidth issues that could *easily* be solved with a little algorithm.
Clever idea. It is the same as building a speaker; the ports (passages) are tuned to only allow through the low frequencies, which the computer does not create.
But will they *finally* use the Netscape/Mozilla browser for the service? If not, that would be a very interesting end to the browser wars: Netscape internet service acessiable only in IE.
It will also add to the tech support confusion as people will no longer understand the distinction between a browser and a service.
I have a DEC VT220 terminal from 1986 connected to my linux server as a console. The plastic has faded to even an unglier color yellow then the terminal orginally was, but it still works perfectly and serves its purpose as well as any replacement would.
In the James Bond movie, The World is Not Enough, a PDA is used when trying to defuse the bomb -- the PDA a disproportinally large Windows logo on the screen and the PDA was held such that the logo was the center of attention rather then what they were doing with the PDA.
It is one thing to have unobtrusive product placement, but I felt that in this scene it was very poorly done -- it was too glaring.
I'm not looking forward to such future product placements.
Same here -- listen to parent. I have a htp374 that used to lock up randomly under heavy load with no error messages. After pulling my hair out for six months trying every debug routine that I could think of I finally put extra space in between the drives for the hell of it -- voila, to my surprise that fixed it. It has been rock solid ever since.
I wanted to swap to VoIP years ago but the powers that be have made it economically infeasible.
If I drop Comcast telephone service and keep Comcast Broadband, the Broadband price nearly goes up 75% the cost of the telephone service (due to their "bundling" prices), so I only save a few dollars which is less then I would save if I swapped to VoIP and purchased a telephone number for incoming calls (as a side note, what it odd about this bundling is that they have not and will not put the telephone and broadband on the same bill -- I still get two bills a month).
OK, then go over to the competition, you say. Well, Verizon will not sell me DSL unless I buy their telephone service and moreover, no one can sell me DSL unless I get Verizon telephone service becuase Verizon owns the lines and will not loan them to third parties for DSL unless I get a telecommuinication service from Verizon to handle the cost of "maintaining the lines" under the 1996 telecommunications act.
So what about SDSL, T1 or leased lines, you say? Not for double to quardruple the price.
Thus, no matter where I go I pretty much *have* to buy telephone service. If it is like this anywhere else the states have nothing to worry about as it will not be economically reasonable to swap to VoIP for residents, at least, and the will get their beloved FCC taxes, which is 37% of the cost of the telephone service itself so they can fund rural health care (what "rural health care" has to do with telephones and the FCC, I have no idea, but a good portion of the FCC line charge go towards that).
Driving non-agressively in my 1997 2.2L standard 2-door Dodge Neon I get 41 highway and 33 city -- and the car was very cheap to buy, there are no modifications and the car has 132hp and a.34 CD (not spectacular at all).
In the end, the Prius, for example, can only do at most 10% better then that so it does not justify twice the cost. I bet if I made my Neon as aerodynamic as the Prius (.26 CD) and put on the same tires (less rolling resistance) I could get near the same gas mileage as the Prius.
Gas mileage is also *heavily* dependant on driving style due to the laws of conservation of energy -- the top things overall that I found increase it are:
1) Standard transmission -- almost all cars with a
standard get better gas mileage then an
automatic. 2) Drive 55 -- Going from 75mph on the highways to
55 increased gas mileage by 26%! 3) Anticipate lights (let the car roll to slow
down when you see a red light ahead of time
rather then breaking at the last minute) saves
a lot, too. 4) Coast down hills 5) Accelerate slowly
If you follow these rules and buy a compact car you'll get near the hybrid's gas mileage with no fancy technology.
"And in the US, a car has become so much part of the identity of "being American", that people would consider even inspecting the system an attack on their civil liberties."
Funny. ..I have to get my car inspected every year for emissions. . .
Not that I agree with it, but in Japan their cars make a beeping or dinging noise if you exceed the maximum speed of the highways -- it lets you pass or go fast for a while for safety but annoys the driver enough to slow down.
If there were to be *any* speed enforcement, I think that this is the best way.
As for "tell the car not to exceed 100kph or whatever, and a simple rev-limiter/electronic throttle would maintain the speed", almost all busses have what they call a governor that if you exceed 65mph, the gas petal slowly puts pressure back up onto the foot until the bus is under 65mph again. This allows for enough time to pass but does not let one speed continuously. If you have a very strong foot you could, however, still potentially speed.
Lastly, as for "I was in a car once where the computer malfunctioned and the fuel-injector was locked full-on. This is equivalent to flooring the accelerator." -- I'm curious as how this happened. I'm not arguing, but what happened from a technical standpoint (this is slashdot afterall)? Basically in most cars the gas pedal is connected to the throttle body which regulates air intake and also has a "TPS" (Throttle position sensor) which sends information to the computer about where the gas petal is and that combined with the mass air flow sensor, the O2 sensor and the knock sensor determine the pulse width of the injectors and thus the gas inflow.
Let's say that something goes haywire and the injectors get stuck full-on. You let go of the gas and the trottle body will close (since it is mechanical) cutting off air to the engine causing it to get flooded for lack of air and stall -- not rev uncontrollably. There is a circumstance, however, where it will rev uncontrollably and that is when the cable to the throttle body gets stuck or the intake manifold cracks when the TPS sensor is fubar'ed. In both cases it is mechanical failure and not electircal and can be solved by turning off the ignition or putting the car in neutral. ..
For the sake of discussion, what are we talking about here?
* Romex, tube-and-knob or BX cabling? * Are there breakers on the mains or is it just
the "big things with electrical tape" you are
talking about? * How is the box wired -- in other words, how did
you trip all the brakers on one side? Generally
breaker boxes are wired back-and-forth with
every other braker on one side wired to the two
incoming 3-pahse mains. This is both for load
balancing accross the mains and also so that you
may put one double breaker accross and get
3-phase 220V. * How many amps is the service? How many circuits
are there?
In the end, there is so little info here that I suggest the same -- call an electrician. It is like someone saying "my computer crashed" with very little info about it or why and asking for advice.
Slashdot Mirror
We did 6" of exterior rockwool, triple pane windows with storms. GREATly reduces outside noise.
http://www.greenbuildingadviso...
It is not cost-effective immediately, however adding that much insulation on the house has about a 15-year payback on heating an cooling costs so it is cost-effective in the long run. Rockwool makes the house fireproof too.
I work for Akamai; Akamai does offer a general-transport better-than-BGP service called Sure Route IP.
.BGP routes based on the least number of hops. . .there are many problems inherant in that. We route based on ping data, bandwidth, cost, reliability, etc, etc, etc.
The idea is that we utilize the massive amounts of data about the Internet's health and the insanely scalable alogorithms for matching end-users to the HTTP server that can best serve them (called mapping) to create generalized IP tunnels that send traffic across "routes" that know more about the Internet then BGP does.
Think about it. .
Did I mention that we are hiring like crazy?
Damn. . .every program I hear on WBUR (NPR) has been showing up on slashdot three days later. . .
Don't worry. . .there is new legislation on the president's desk right now that will make it "virtually impossible now to get a nationwide class-action suit off the ground." -- The Economist
> The only reason I use Firefox (and I use it a
.you just have to write a proxy.pac (pac = proxy auto-config). It is a javascript file that points your browser to the proper proxy server depending on the request.
/ de mo/proxy-live.html
> lot) is that I can't split proxy servers in
> Opera.
Sure you can. .
For reference:
http://wp.netscape.com/eng/mozilla/2.0/relnotes
http://www.fas.org/man/dod-101/sys/ac/c-5.htm
No, you're right. They actually do. At the college I used to go to, for example, they hired and paid a student to be a "Microsoft Evangellist" (they actually titled the position that) whose sole job is to say that Microsoft is great.
Read Karl Marx's Kapital -- specifically "The Labor Theory of Value". He architects beautifully the argument you just made.
Here's a solution to the whole debate -- make the sticky bit have meaning under Linux like it does on other UNIXen -- if the sticky bit is set on the execuatble, do not swap it. If it is not set, the executable is free to be swapped. This solves the entire debate (for instance, if you don't want the 'interactive' mozilla process swapped, set the sticky bit on the executable).
The G400 eTV is a nice card:
see here
Drawbacks:
The nicest thing about the card is that specs for most of the chips are avaiable if you wish to hack it.
To see more about the linux support:
See here.
There are many different implementations of QoS and some solve different problems:
* Some give priority to specific protocols or
ports
* Some give priority to smaller packets over
larger
* Some give priority based on the ToS bit
* Some just drop packets for "high bandwidth"
connections and hope IP drops its window size
There are also about a dozen different ways to do each of the above and each vendor has a different name for it. Thus there is much confusion on this subject and I did not make it clear in my post which I was speaking of.
The idea is to schedule packets based upon origin / destination IP address rather then by a protocol, port or connection. The simplest schemes such as WRR mentioned above, require little CPU because they are simply round-robin. The down-side to this is that they make the TCP window size thrash. More advanced algorithms use a modified token-bucket scheme and grant a specific number of tokens per second to each IP address it sees.
In both cases, because it is done by IP and not by protocol or connection, Joe P Hacker can have 1,000 P2P connections going and if Grandma loads a web page, Joe's connections will not slow her down while still giving Joe all the availible bandwidth he wants. If Grandma and Sue down the street load a web page, again they will load fast (slowing down Joe for a second) and then Joe resumes. If Joe loads a web page while he has 1,000 P2P connections, his own connections will drown him out. If Joe P Hacker is running 1,000 P2P connections while Julie is downloading the latest Linux distro and they are the only two on the wire, they'll get equal bandwidth.
Who cares if the pipe is saturated by P2P people if we can guarantee that everyone else's traffic gets through when they need it?
So this type of a scheme is not a backbone solution, it is a near-leaf solution. ISPs implement this scheme (or at least should) within regions and then balance their regional routers with another scheme more suited to massive bandwidth.
Granted, I have never done this in a NAP scenario, however in a company with ~500 employees after proper tuning I never had to worry about any one person downloading too much crap ever again. But as mentioned before, this would not be done at the NAP anyway -- it should be done more towards the leaves.
Note that I said there are other implementations that scale better in the original post, and that WRR was just an example to illustrate the point.
There are many papers on this subject and many routers implement the protocol -- even to "hundreds of thousands of hosts".
Google for "Weighted Fair Queueing". I don't have the time right now to right a tutorial on the subject. There are papers that answer your other questions.
The solution ISPs can use is weighted fair queueing.
This ensures that one person does not destroy the bandwidth of another. It is a hell of a lot better then making users worry about how much they download.
One such implementation is the Weighted Round Robin qdisc in Linux:
http://wipl-wrr.dkik.dk/wrr/
There are other implementations that scale better.
I say this every time someone brings up the "scarce bandwidth" issue, but no one ever listens and ISP continue to use draconian way to solve their bandwidth issues that could *easily* be solved with a little algorithm.
Clever idea. It is the same as building a speaker; the ports (passages) are tuned to only allow through the low frequencies, which the computer does not create.
But will they *finally* use the Netscape/Mozilla browser for the service? If not, that would be a very interesting end to the browser wars: Netscape internet service acessiable only in IE.
It will also add to the tech support confusion as people will no longer understand the distinction between a browser and a service.
I have a DEC VT220 terminal from 1986 connected to my linux server as a console. The plastic has faded to even an unglier color yellow then the terminal orginally was, but it still works perfectly and serves its purpose as well as any replacement would.
They've already been trying it.
In the James Bond movie, The World is Not Enough,
a PDA is used when trying to defuse the bomb -- the PDA a disproportinally large Windows logo on the screen and the PDA was held such that the logo was the center of attention rather then what they were doing with the PDA.
It is one thing to have unobtrusive product placement, but I felt that in this scene it was very poorly done -- it was too glaring.
I'm not looking forward to such future product placements.
Same here -- listen to parent. I have a htp374 that used to lock up randomly under heavy load with no error messages. After pulling my hair out for six months trying every debug routine that I could think of I finally put extra space in between the drives for the hell of it -- voila, to my surprise that fixed it. It has been rock solid ever since.
I wanted to swap to VoIP years ago but the
powers that be have made it economically infeasible.
If I drop Comcast telephone service and keep Comcast Broadband, the Broadband price nearly goes up 75% the cost of the telephone service (due to their "bundling" prices), so I only save a few dollars which is less then I would save if I swapped to VoIP and purchased a telephone number for incoming calls (as a side note, what it odd about this bundling is that they have not and will not put the telephone and broadband on the same bill -- I still get two bills a month).
OK, then go over to the competition, you say. Well, Verizon will not sell me DSL unless I buy their telephone service and moreover, no one can sell me DSL unless I get Verizon telephone service becuase Verizon owns the lines and will not loan them to third parties for DSL unless I get a telecommuinication service from Verizon to handle the cost of "maintaining the lines" under the 1996 telecommunications act.
So what about SDSL, T1 or leased lines, you say? Not for double to quardruple the price.
Thus, no matter where I go I pretty much *have* to buy telephone service. If it is like this anywhere else the states have nothing to worry about as it will not be economically reasonable to swap to VoIP for residents, at least, and the will get their beloved FCC taxes, which is 37% of the cost of the telephone service itself so they can fund rural health care (what "rural health care" has to do with telephones and the FCC, I have no idea, but a good portion of the FCC line charge go towards that).
I've wondered that too; but there is a solution if
you compile statically with these flags on gcc:
export CFLAGS="-ffunction-sections -fdata-sections"
export LDFLAGS="-static -Wl,--gc-sections"
On both the libraries and the code that links them.
Driving non-agressively in my 1997 2.2L standard .34 CD (not spectacular at all).
2-door Dodge Neon I get 41 highway and 33 city -- and the car was very cheap to buy, there are no modifications and the car has 132hp and a
In the end, the Prius, for example, can only do at most 10% better then that so it does not justify twice the cost. I bet if I made my Neon as aerodynamic as the Prius (.26 CD) and put on the same tires (less rolling resistance) I could get near the same gas mileage as the Prius.
Gas mileage is also *heavily* dependant on driving style due to the laws of conservation of energy -- the top things overall that I found increase it are:
1) Standard transmission -- almost all cars with a
standard get better gas mileage then an
automatic.
2) Drive 55 -- Going from 75mph on the highways to
55 increased gas mileage by 26%!
3) Anticipate lights (let the car roll to slow
down when you see a red light ahead of time
rather then breaking at the last minute) saves
a lot, too.
4) Coast down hills
5) Accelerate slowly
If you follow these rules and buy a compact car you'll get near the hybrid's gas mileage with no fancy technology.
"And in the US, a car has become so much part of the identity of "being American", that people would consider even inspecting the system an attack on their civil liberties."
.I have to get my car inspected every year for emissions. . .
Funny. .
Not that I agree with it, but in Japan their cars make a beeping or dinging noise if you exceed the maximum speed of the highways -- it lets you pass or go fast for a while for safety but annoys the driver enough to slow down.
.
If there were to be *any* speed enforcement, I think that this is the best way.
As for "tell the car not to exceed 100kph or whatever, and a simple rev-limiter/electronic throttle would maintain the speed", almost all busses have what they call a governor that if you exceed 65mph, the gas petal slowly puts pressure back up onto the foot until the bus is under 65mph again. This allows for enough time to pass but does not let one speed continuously. If you have a very strong foot you could, however, still potentially speed.
Lastly, as for "I was in a car once where the computer malfunctioned and the fuel-injector was locked full-on. This is equivalent to flooring the accelerator." -- I'm curious as how this happened. I'm not arguing, but what happened from a technical standpoint (this is slashdot afterall)? Basically in most cars the gas pedal is connected to the throttle body which regulates air intake and also has a "TPS" (Throttle position sensor) which sends information to the computer about where the gas petal is and that combined with the mass air flow sensor, the O2 sensor and the knock sensor determine the pulse width of the injectors and thus the gas inflow.
Let's say that something goes haywire and the injectors get stuck full-on. You let go of the gas and the trottle body will close (since it is mechanical) cutting off air to the engine causing it to get flooded for lack of air and stall -- not rev uncontrollably. There is a circumstance, however, where it will rev uncontrollably and that is when the cable to the throttle body gets stuck or the intake manifold cracks when the TPS sensor is fubar'ed. In both cases it is mechanical failure and not electircal and can be solved by turning off the ignition or putting the car in neutral. .
Why was there an outage in Boston, MA and Springfield, MA -- Massachusetts did not loose any power?!
For the sake of discussion, what are we talking about here?
* Romex, tube-and-knob or BX cabling?
* Are there breakers on the mains or is it just
the "big things with electrical tape" you are
talking about?
* How is the box wired -- in other words, how did
you trip all the brakers on one side? Generally
breaker boxes are wired back-and-forth with
every other braker on one side wired to the two
incoming 3-pahse mains. This is both for load
balancing accross the mains and also so that you
may put one double breaker accross and get
3-phase 220V.
* How many amps is the service? How many circuits
are there?
In the end, there is so little info here that I suggest the same -- call an electrician. It is like someone saying "my computer crashed" with very little info about it or why and asking for advice.