You are vastly overestimating the amount of bandwidth a typical ADSL DSLAM gets. There are 300-port DSLAMs hooked up to a single DS3 out there. A 24-port would _maybe_ have a couple of 1.5Mbit HDSL lines going to it. Telcos sure as hell don't plan on residential users all using BitTorrent at the same time. The 80/20 rule certainly applies here.
The smaller DSLAMs might even be useful in residential areas. Uptake rate for ADSL isn't exactly high, especially when many of your customers can only get really slow DSL because they are too far from the CO. I'd imagine there are plenty of smaller DSLAMs being used in low-density residential neighborhoods which are too far from the CO but still have plenty of DSL customers. You can't really put a huge pedestal in that application, there is often very little space available.
You didn't read the article very carefully. You can use multiple wire pairs to get more power. This ups your maximum wattage quite a bit. I think a 48-port OSP DSLAM uses about 50 watts, and someone must be buying the damn things. I am not saying Bellsouth does, but they are definitely a very nice solution for middle-of-nowhere applications where you can't afford to splurge on a pedestal when you only have 50-100 customers in a 10-mile radius. You'll start seeing a lot of that with FTTN -- VDSL can't go very far at all, so each neighborhood will need a small DSLAM.
Well, their quality really declined when they were still making money -- in the 80s. It got even worse in the 90s. After their cars got a reputation of being consistently unreliable, Toyota and Honda pretty much took over that market. GM was happy while it could still make expensive SUVs with a huge profit margin (and sell most of its cars to fleet customers), but once the high fuel prices hit, they were basically fucked. Of course, now they have no cars that sell, a bad reputation among 90% of consumers, and union contracts that prevent them from downsizing as fast as they would like. This situation is hard to get out of without losing a ton of money. They'll probably come back, just cause all those investors pounding their ass is a real wake-up call for their idiot management.
WTF dude. Ardour in no way, shape, or form compares to Pro Tools. It can't even do everything Cubase can do. There's a reason why almost every professional studio uses ProTools. I've never heard of a single one using Ardour. I'm sure it will be a neat little package a few years down the road, but it just doesn't compare favorably to a $50,000 one.
Well, they are assembled in the USA. Most of the parts are made in Japan or Mexico. The issue is not really where the work is done, it's more about how much the company cares about quality. The Japanese have an obsession with quality, they invented a lot of things in that area. Quality systems like ISO9001 and methods such as lean manufacturing are all borrowed from Japanese manufacturers. Companies like GM are happy if their cars have an order of magnitude more production problems than Toyota, as long as they make money. That's why they are currently getting whipped by the marketplace.
Ping times don't mean anything if the connection isn't loaded. Try doing the same ping when you have call traffic. Many routers and cable modems shit themselves when they have to route many small packets.
Uh, it's not Comcast, it's more the "heavy uploading". Cable is pretty slow upstream, and you probably only have 256 Kbps total upstream. Voice requires at least 128 Kbps. If you are doing ANYTHING with the connection, you won't have enough bandwidth left over to ensure a low latency connection. Packets will get queued up in your cable modem, and you will get dropouts. To get decent quality, you'd have to completely stop all other internet traffic, and even then you'd probably get occasional dropouts.
Well, it's nobody but your own damn fault for buying a house in an HOA. Next thing you know, they'll be billing you for maintenance on a nonexistent pool and fining you for setting your trash out on trash day. An HOA is about the closest you can get to legal organized crime.
Unless you are in outer mongolia, ISPs most certainly do not use satellites. Satellite internet is slow, expensive, and has very high latency (it takes a quarter of a second for data just to bounce up there and back). ISPs generally use fiberoptic connections, usually something like an OC-3 (or a regular coax T3 if they are a small ISP).
Incorrect. DSLAMs use more power than can be (safely) run down a 26AWG pair of a 25 or 50 pair trunk.
There are plenty of DSLAMs that are powered from line pairs. They might not use ONE pair, they can use several. A DSLAM is pretty small, they wouldn't mount just one on a pedestal with its own power meter. You can buy DSLAMs that are about the size of a 19" monitor and mount directly onto a pole. The RTs you have seen probably have tons of other things as well (fiber muxes, T1 repeaters, HDSL units, ISDN equipment, etc).
Nearly all telco equipment that's outside runs off of line power. Putting in pedestals with their own rectifiers is really fucking expensive and is only done in high-density metro areas. There are tons of T1 repeaters mounted inside manholes and such, do you think they'll have a rectifier inside a manhole?
DSL is a bit of a challenge, too. However, DSL can be monitored from other pairs in the same trunk just like a cable modem. (listen for weak cross-talk.)
That's one of the funniest things I've ever heard. It's a major bitch to decode DSL if you are on the other end of a long pair -- you need a pretty good front end to get rid of the ISI, echo, and other channel problems. There is no way in hell you could decode DSL just listening to crosstalk. For a start, how would you tell one line from another?
The amplifiers are line powered. The cable trunk has a few hundred volts DC across it. Not to mention, cable companies often have UPSs installed on poles. Ever see the big gray boxes that say "Alpha" on them? Those are UPSes.
Where I live, the phone lines are buried. I suppose you could get into a manhole somewhere and listen in, but how the hell would you cut into the trunk? Not to mention, if you get caught, better get ready for some ass-pounding in prison. Same thing applies for screwing with the NID or climbing poles outside my house. And with a cable modem, anyone on your segment can look at your traffic; it's a shared medium.
Look dude, I work for a pretty large telecom supplier. I know how this shit works, OK? A BRI (basic rate ISDN) line is 64 Kbps. However, this is a CIRCUIT SWITCHED LINE. It's not a "maximum" of 64 Kbps, it's exactly 64 Kbps. When it's used for voice, you transmit one raw sample every 125 microseconds. There is no need for buffering anything. Your latency is practically zero.
A cable or DSL connection is a PACKET SWITCHED line. You can't transmit raw samples of voice, they have to be buffered in a packet. If you want acceptable latency, you have to send out very small packets, where the packet overhead is larger than the data payload. This means a "64 Kbit" codec will use anywhere from 128 Kbps to 350 Kbps depending on how low you want the latency to be. Even then, the latency is more than 100 times higher than on a phone line JUST TO BUFFER THE PACKET. Transmitting it takes even longer. If your connection isn't extremely low latency, you'll get really shitty audio even if you get the advertised bandwidth. Using increased compression isn't particularly effective, since packet overhead tends to dominate.
I really doubt the cable company is doing anything to your packets. What is far more likely is that an increase in the number of ISP subscribers increases your latency to the point where VoIP doesn't work that well. Since you don't have latency guarantees, this is something you just have to deal with -- or pay for a connection with guaranteed latency. Network neutrality laws will just eliminate the second option, and Vonage won't be any better-off.
Bullshit. You obviously have no goddamn clue about how networks work. Web browsing demands maybe 250 ms of latency. Voice demands at least 10 times less. Video is even more demanding. Oh, and that 64 Kbps is only for an unpacketized, dedicated DS0. With the packet overhead, it's more like 350 Kbps. It's not the bitrate that matters, it's the latency and jitter. Nobody wants a voice packet if it gets there 20 milliseconds late. There's a reason you don't need echo cancellers on regular phone lines: there's hardly any latency compared to even VoIP on a gigabit LAN.
No, you don't. You don't pay for a "voice-grade connection". You pay for a connection with, say, 3Mbits down and 256 Kbps up. Show me where it says "and you get less than 50 ms latency". I'm sure you get pretty close to the advertised bitrates when you download a file from Vonage's website. If you want to pay extra for lower latency and guaranteed levels of packet loss, you should have the ability to do that. It certainly isn't free for the ISP to provide this higher grade of service.
The other way of looking at it is that at that point, I'm no longer paying for internet service, but rather some subset that is defined by Time Warner's priorities.
Internet service has never included real-time guarantees. A good connection means that your packets make it to their destination. It doesn't guarantee that this happens within a certain time period.
I agree. My suspicions are that Vonage has gained customers faster than it has expanded its infrastructure. I have nothing specific to back that up.
Well, it's also possible that Time Warner's network has increased the number of subscribers. This will also cause the latency to increase, even if bitrates do not significantly decrease. If you have 1.5Mbps of traffic and a 1.5Mbps pipe your latency will be a lot higher than if you have 1.5Mbps of traffic and a 30 Mbps pipe.
Vonage has no QoS. The Internet in general has no QoS. The sound quality adjustment simply adjusts how much it compresses the data. The prioritization only works on your part of the network. Once it gets onto the internet, it's a free-for-all.
Well, maybe you wouldn't have this problem if Vonage spent the money to build their own damn network instead of counting on cable companies spending the cash to build a voice-grade network that they can use for free. Your agreement with Time Warner doesn't guarantee your connection is voice-grade, so why should they have the obligation to providing such a connection? Low latency requires spending lots of money to increase network capacity, so why should you or Vonage get this extra benefit for free? Net neutrality advocates seem to forget that network capacity costs real money, and that everyone can't have low latency unless the network has lots of unused capacity. Your connection quality probably went down due to an increase in the number of subscribers causing your latency to increase, not because Time Warner is purposefully screwing with Vonage.
Well, they aren't really lying. Vonage often has poor call quality, since there is no QoS. Vonage is not secure, since no encryption is used and the packets travel on a public network. The only real bullshit claim is the PC thing, and even then it's partially true (if PPPoE is used and the Vonage box is not configured appropriately, it will not work).
2. It's really a new service by "Yahool Mail Beta" and not "Yahoo! Mail Beta". Yahool is a Trademark of Google Inc. and is not to be confused with Yahoo! in any way!
You're right, I think it must have been around 40,000 hours. I just remembered that it was close to a round number. The drive was one of the first 100GB IDE drives out there, and it was running pretty much 24/7.
And of course, the Linux kernel does not, and never has, required patches to be submitted before they're used.
Yeah, and that's part of the reason why Linux will probably never be a single, unified platform. Every linux distro has significant differences in how it does things, and this is a very bad thing. In fact, this is why Unix pretty much died as a platform and Linux is slowly sliding down the same slope. The #1 complaint against Linux is lack of software, and the reason is obvious -- the code often needs to be hacked up to run on a given distribution.
Unlike Debian, Firefox actually aims to be a mainstream, successful open-source product. Firefox has several orders of magnitude more users than any other open-source piece of software. To remain in that spot, Firefox releases need to be of high quality, standardized to ensure compatibility with sites and extensions, and have consistent branding and marketing. When Debian's developers fuck with the codebase and introduce new bugs, it's Mozilla's reputation that gets tarnished -- even though Debian screwed it up. Furthermore, bug reports are not very useful when users aren't using a standard release.
Debian is being extremely rude and egotistical here. If they don't want to ship an unmolested version of Firefox, they shouldn't include it at all. If they feel the need to make changes to Firefox, they should follow Mozilla's development process. It's not their project to fuck with. If they want to hack shit onto it without following Mozilla's development process, they need to fork the codebase and use a different name. When millions of people start using Debian, then they can bitch about Mozilla's branding policy. Until then, maybe projects like Debian should look at why hardly anyone uses them before they start whining about Firefox.
Nah, there's no trend there. I've actually had far fewer problems with my 100 and 160 gig drives than I had with the ~10-20 gig stuff. I had only one 100 gig drive fail, and according to SMART it had about a hundred thousand hours on the clock, which was way past its expected lifetime. Even then, it didn't fail completely, just got extremely slow. I was able to copy all the data off of it, although that took about 15 hours. Western Digital seems to make pretty damn good drive controllers.
Spinup/spindown are actually not significant as far as wear is concerned. An otherwise worn-out drive is more likely to fail during a spin-up or spin-down, but those actions don't actually cause any wear. My understanding is that most failures are caused by spindle bearings wearing out, which is directly proportional to how long the drive has been powered on.
Slashdot Mirror
You are vastly overestimating the amount of bandwidth a typical ADSL DSLAM gets. There are 300-port DSLAMs hooked up to a single DS3 out there. A 24-port would _maybe_ have a couple of 1.5Mbit HDSL lines going to it. Telcos sure as hell don't plan on residential users all using BitTorrent at the same time. The 80/20 rule certainly applies here.
The smaller DSLAMs might even be useful in residential areas. Uptake rate for ADSL isn't exactly high, especially when many of your customers can only get really slow DSL because they are too far from the CO. I'd imagine there are plenty of smaller DSLAMs being used in low-density residential neighborhoods which are too far from the CO but still have plenty of DSL customers. You can't really put a huge pedestal in that application, there is often very little space available.
You didn't read the article very carefully. You can use multiple wire pairs to get more power. This ups your maximum wattage quite a bit. I think a 48-port OSP DSLAM uses about 50 watts, and someone must be buying the damn things. I am not saying Bellsouth does, but they are definitely a very nice solution for middle-of-nowhere applications where you can't afford to splurge on a pedestal when you only have 50-100 customers in a 10-mile radius. You'll start seeing a lot of that with FTTN -- VDSL can't go very far at all, so each neighborhood will need a small DSLAM.
Well, their quality really declined when they were still making money -- in the 80s. It got even worse in the 90s. After their cars got a reputation of being consistently unreliable, Toyota and Honda pretty much took over that market. GM was happy while it could still make expensive SUVs with a huge profit margin (and sell most of its cars to fleet customers), but once the high fuel prices hit, they were basically fucked. Of course, now they have no cars that sell, a bad reputation among 90% of consumers, and union contracts that prevent them from downsizing as fast as they would like. This situation is hard to get out of without losing a ton of money. They'll probably come back, just cause all those investors pounding their ass is a real wake-up call for their idiot management.
WTF dude. Ardour in no way, shape, or form compares to Pro Tools. It can't even do everything Cubase can do. There's a reason why almost every professional studio uses ProTools. I've never heard of a single one using Ardour. I'm sure it will be a neat little package a few years down the road, but it just doesn't compare favorably to a $50,000 one.
Well, they are assembled in the USA. Most of the parts are made in Japan or Mexico. The issue is not really where the work is done, it's more about how much the company cares about quality. The Japanese have an obsession with quality, they invented a lot of things in that area. Quality systems like ISO9001 and methods such as lean manufacturing are all borrowed from Japanese manufacturers. Companies like GM are happy if their cars have an order of magnitude more production problems than Toyota, as long as they make money. That's why they are currently getting whipped by the marketplace.
Ping times don't mean anything if the connection isn't loaded. Try doing the same ping when you have call traffic. Many routers and cable modems shit themselves when they have to route many small packets.
Uh, it's not Comcast, it's more the "heavy uploading". Cable is pretty slow upstream, and you probably only have 256 Kbps total upstream. Voice requires at least 128 Kbps. If you are doing ANYTHING with the connection, you won't have enough bandwidth left over to ensure a low latency connection. Packets will get queued up in your cable modem, and you will get dropouts. To get decent quality, you'd have to completely stop all other internet traffic, and even then you'd probably get occasional dropouts.
Well, it's nobody but your own damn fault for buying a house in an HOA. Next thing you know, they'll be billing you for maintenance on a nonexistent pool and fining you for setting your trash out on trash day. An HOA is about the closest you can get to legal organized crime.
Unless you are in outer mongolia, ISPs most certainly do not use satellites. Satellite internet is slow, expensive, and has very high latency (it takes a quarter of a second for data just to bounce up there and back). ISPs generally use fiberoptic connections, usually something like an OC-3 (or a regular coax T3 if they are a small ISP).
Incorrect. DSLAMs use more power than can be (safely) run down a 26AWG pair of a 25 or 50 pair trunk.
There are plenty of DSLAMs that are powered from line pairs. They might not use ONE pair, they can use several. A DSLAM is pretty small, they wouldn't mount just one on a pedestal with its own power meter. You can buy DSLAMs that are about the size of a 19" monitor and mount directly onto a pole. The RTs you have seen probably have tons of other things as well (fiber muxes, T1 repeaters, HDSL units, ISDN equipment, etc).
Nearly all telco equipment that's outside runs off of line power. Putting in pedestals with their own rectifiers is really fucking expensive and is only done in high-density metro areas. There are tons of T1 repeaters mounted inside manholes and such, do you think they'll have a rectifier inside a manhole?
DSL is a bit of a challenge, too. However, DSL can be monitored from other pairs in the same trunk just like a cable modem. (listen for weak cross-talk.)
That's one of the funniest things I've ever heard. It's a major bitch to decode DSL if you are on the other end of a long pair -- you need a pretty good front end to get rid of the ISI, echo, and other channel problems. There is no way in hell you could decode DSL just listening to crosstalk. For a start, how would you tell one line from another?
The amplifiers are line powered. The cable trunk has a few hundred volts DC across it. Not to mention, cable companies often have UPSs installed on poles. Ever see the big gray boxes that say "Alpha" on them? Those are UPSes.
I always thought the standard for TDM lines was 99.99% and for packet it was 99.95%.
Where I live, the phone lines are buried. I suppose you could get into a manhole somewhere and listen in, but how the hell would you cut into the trunk? Not to mention, if you get caught, better get ready for some ass-pounding in prison. Same thing applies for screwing with the NID or climbing poles outside my house. And with a cable modem, anyone on your segment can look at your traffic; it's a shared medium.
Look dude, I work for a pretty large telecom supplier. I know how this shit works, OK? A BRI (basic rate ISDN) line is 64 Kbps. However, this is a CIRCUIT SWITCHED LINE. It's not a "maximum" of 64 Kbps, it's exactly 64 Kbps. When it's used for voice, you transmit one raw sample every 125 microseconds. There is no need for buffering anything. Your latency is practically zero.
A cable or DSL connection is a PACKET SWITCHED line. You can't transmit raw samples of voice, they have to be buffered in a packet. If you want acceptable latency, you have to send out very small packets, where the packet overhead is larger than the data payload. This means a "64 Kbit" codec will use anywhere from 128 Kbps to 350 Kbps depending on how low you want the latency to be. Even then, the latency is more than 100 times higher than on a phone line JUST TO BUFFER THE PACKET. Transmitting it takes even longer. If your connection isn't extremely low latency, you'll get really shitty audio even if you get the advertised bandwidth. Using increased compression isn't particularly effective, since packet overhead tends to dominate.
I really doubt the cable company is doing anything to your packets. What is far more likely is that an increase in the number of ISP subscribers increases your latency to the point where VoIP doesn't work that well. Since you don't have latency guarantees, this is something you just have to deal with -- or pay for a connection with guaranteed latency. Network neutrality laws will just eliminate the second option, and Vonage won't be any better-off.
Bullshit. You obviously have no goddamn clue about how networks work. Web browsing demands maybe 250 ms of latency. Voice demands at least 10 times less. Video is even more demanding. Oh, and that 64 Kbps is only for an unpacketized, dedicated DS0. With the packet overhead, it's more like 350 Kbps. It's not the bitrate that matters, it's the latency and jitter. Nobody wants a voice packet if it gets there 20 milliseconds late. There's a reason you don't need echo cancellers on regular phone lines: there's hardly any latency compared to even VoIP on a gigabit LAN.
I pay for my network connection.
No, you don't. You don't pay for a "voice-grade connection". You pay for a connection with, say, 3Mbits down and 256 Kbps up. Show me where it says "and you get less than 50 ms latency". I'm sure you get pretty close to the advertised bitrates when you download a file from Vonage's website. If you want to pay extra for lower latency and guaranteed levels of packet loss, you should have the ability to do that. It certainly isn't free for the ISP to provide this higher grade of service.
The other way of looking at it is that at that point, I'm no longer paying for internet service, but rather some subset that is defined by Time Warner's priorities.
Internet service has never included real-time guarantees. A good connection means that your packets make it to their destination. It doesn't guarantee that this happens within a certain time period.
I agree. My suspicions are that Vonage has gained customers faster than it has expanded its infrastructure. I have nothing specific to back that up.
Well, it's also possible that Time Warner's network has increased the number of subscribers. This will also cause the latency to increase, even if bitrates do not significantly decrease. If you have 1.5Mbps of traffic and a 1.5Mbps pipe your latency will be a lot higher than if you have 1.5Mbps of traffic and a 30 Mbps pipe.
Vonage has no QoS. The Internet in general has no QoS. The sound quality adjustment simply adjusts how much it compresses the data. The prioritization only works on your part of the network. Once it gets onto the internet, it's a free-for-all.
Well, maybe you wouldn't have this problem if Vonage spent the money to build their own damn network instead of counting on cable companies spending the cash to build a voice-grade network that they can use for free. Your agreement with Time Warner doesn't guarantee your connection is voice-grade, so why should they have the obligation to providing such a connection? Low latency requires spending lots of money to increase network capacity, so why should you or Vonage get this extra benefit for free? Net neutrality advocates seem to forget that network capacity costs real money, and that everyone can't have low latency unless the network has lots of unused capacity. Your connection quality probably went down due to an increase in the number of subscribers causing your latency to increase, not because Time Warner is purposefully screwing with Vonage.
Well, they aren't really lying. Vonage often has poor call quality, since there is no QoS. Vonage is not secure, since no encryption is used and the packets travel on a public network. The only real bullshit claim is the PC thing, and even then it's partially true (if PPPoE is used and the Vonage box is not configured appropriately, it will not work).
What are you smoking?
You're right, I think it must have been around 40,000 hours. I just remembered that it was close to a round number. The drive was one of the first 100GB IDE drives out there, and it was running pretty much 24/7.
And of course, the Linux kernel does not, and never has, required patches to be submitted before they're used.
Yeah, and that's part of the reason why Linux will probably never be a single, unified platform. Every linux distro has significant differences in how it does things, and this is a very bad thing. In fact, this is why Unix pretty much died as a platform and Linux is slowly sliding down the same slope. The #1 complaint against Linux is lack of software, and the reason is obvious -- the code often needs to be hacked up to run on a given distribution.
Unlike Debian, Firefox actually aims to be a mainstream, successful open-source product. Firefox has several orders of magnitude more users than any other open-source piece of software. To remain in that spot, Firefox releases need to be of high quality, standardized to ensure compatibility with sites and extensions, and have consistent branding and marketing. When Debian's developers fuck with the codebase and introduce new bugs, it's Mozilla's reputation that gets tarnished -- even though Debian screwed it up. Furthermore, bug reports are not very useful when users aren't using a standard release.
Debian is being extremely rude and egotistical here. If they don't want to ship an unmolested version of Firefox, they shouldn't include it at all. If they feel the need to make changes to Firefox, they should follow Mozilla's development process. It's not their project to fuck with. If they want to hack shit onto it without following Mozilla's development process, they need to fork the codebase and use a different name. When millions of people start using Debian, then they can bitch about Mozilla's branding policy. Until then, maybe projects like Debian should look at why hardly anyone uses them before they start whining about Firefox.
Nah, there's no trend there. I've actually had far fewer problems with my 100 and 160 gig drives than I had with the ~10-20 gig stuff. I had only one 100 gig drive fail, and according to SMART it had about a hundred thousand hours on the clock, which was way past its expected lifetime. Even then, it didn't fail completely, just got extremely slow. I was able to copy all the data off of it, although that took about 15 hours. Western Digital seems to make pretty damn good drive controllers.
Spinup/spindown are actually not significant as far as wear is concerned. An otherwise worn-out drive is more likely to fail during a spin-up or spin-down, but those actions don't actually cause any wear. My understanding is that most failures are caused by spindle bearings wearing out, which is directly proportional to how long the drive has been powered on.