You have an interesting definition of "not expensive". Fluke doesn't make anything that's cheap. And the NetTool is a $2500 device placing it well out of the budget of anyone who doesn't mess with cabling a lot.
Answer: Beldin (or whoever made the physical cable)
(I just crimped the ends on it.)
And I don't like the phrase "the cable went bad". Someone or something has to do something to it to make it stop working. A cable that's been plugged in and working for months or years will continue to do so right up to the point someone starts messing with it.
To that I say the people making your cables obviously aren't qualified to be making them.
Cables are either "good" or "bad"; they are either within spec or not. There's no such thing as a "marginal good" cable. I read your last sentence as meaning the cable is within spec as per whatever tested it, but then your network hardware isn't -- or your tester is crap.
Actually, it's a spec thing. The wiring specifications say 100m. So, by definition, any cable longer than 100m is out-of-spec. That said, in the old days, it read "200m node to node" which translates to 100m cables (100m node-switch x 2) And with certain switches (class I vs II) there could be an additional 2m or 5m between switches -- making the total 205m, max. (But that's from memory back when 100base-TX wasn't a standard.)
With the advent of FDX, line length has become less important. And people get away with cables far beyond spec. As long as there's enough signal at both ends.
I cannot speak to the cheap crap sold at Lowes, Home Depot, etc. However, the professional crimps use the same types of blades as the mass produced cables. Those "superior" factory made cables work loose over time, as well. And all those injection molded one's I've ever seen have much smaller actual crimps that make them much weaker. I've never had a cable I made break, but I've had numerous Belkin and Uniden factory made cables come completely apart and "work loose".
1 and 2 are made by humans. 3 is not. And it's very easy to tell them apart... the robots don't use the same crimps as they are injection molded. (and the wires are 100% straight and uniformly pushed to the back of the crimp.) The only way to make money making cables is to use robots. Lots of robots. People are far too expensive, slow, and prone to errors.
Right. The correct solution is to not aim a satelite transmitter through the edge of a building. The working solution is to sheild the T1 line and ground it sufficiently to keep out the RF causing errors. (which in this case was 3 places. admitedly, getting a proper ground on the 3rd floor of a building being baked by a dish is a problem in itself.)
But you go right a head and tell me how to fix problems you've never had to fix within systems where you have zero knowledge, experience, or expertise. "Well, that copper drain pipe should be grounding the flashing so there shouldn't be RF throughout the building."
That was not a simple "parabolic reflector". That was a transmitter for a radio station's uplink to it's transmitter station. So, far more power than most people are aware.
You can go on believing what ever you want. Until you see it for yourself, you are going to stand by what your books and theories tell you. It's people like you that make troubleshooting problems such a pain in the ass as you won't look for things you are certain are "impossible".
And no, putting a ground in one place will not "always" do it for you. Did you miss the part about a satelite transmitter next door? Aimed into the edge of the building? That's a lot of signal to keep out of the system... and it's literally everywhere.
Actually, I prefer my equipment not getting struck by lightning. As such, I use conduits were possible. And go out of my way to not hang stuff up in the air.
(And now that I think about it, all of the damage I've dealt with in the last few years has been at the hands of cable TV connections. 'tho several years ago, I had an ISDN NT-1 blown to bits, along with the cards in the RT and CO *grin* despite the entire 4.7 miles of cable being underground.)
Ah, I see we have another NCSU Alum from the Great Squirrel Power Outage of '90 (maybe 91?) "We know it was a squirrel due to the carbonized foot and tail left on the bus bar." Killed power to the entire main campus for most of the day.
Or you learn to stop plugging lightning rods directly into your switch. Plug it into a patch panel / grounding block instead where it can be properly grounded vs. the tiny metal fingers found in RJ45 ports. (If it's direct strike, it's going to blow up whatever is connected at both ends no matter how well ground.)
... or tightly zip tie large bundles to trusses in the ceiling where you cannot see them or reach them without a crane. I've seen some zipped so tight it's cutting through the outer insulation.
And I can recall one location who's T1 was always flaky. It turned out to be the radio dish next door aimed across the top corner of the building where there was a copper drain pipe and copper flashing. It was enough RF noise to cause signalling issues. A liberal application of sheilded cables and ground straps fixed it up. (and no, there was nothing that could be done about the dish.)
(work in the telecom world long enough and you really can see everything:-))
The orginal article left out Cat3... coax (10b2) -> cat3 (10bT) -> cat5 (100bT) -> cat5e (1000bT) -> cat6 (10Gb, short runs) and cat6a (10Gb, full length) [and I'm leaving out 10b5/"thicknet"]
Your installation sounds like a very old phone wiring (cat3) that was repurposed for ethernet. Cat3 will work for 100bT, up to a point. It will not work very reliablly for 1000bT -- 'tho you may get lucky using cat3 patch cords. The 2 pair installations will not work beyond 100bT as all 4 pairs are used for GigE and 10Gb.
My nightmare is seeing one cable (usually cat3!) carrying *2* ethernets -- in clear disregard for documented cabling standards. And a coworker (network tech) who only crimped 2 pairs per end -- he'd leave the other 2 pairs out of the crimp on purpose... "you don't need those." (he had to carry around three cables... an ethernet cable, a T1 cable, and a serial cable because he wouldn't crimp one cable with all 4 pairs.)
I don't really have any idea what ISPs actually do or how internet infrastructure works.
Magic(tm).
It works just like your ad-hoc "connect to a friend" example. Your friend is then the ISP. He's responsible moving traffic to those who are directly connected to him. All the homes in your neighborhood connect to one house who connects to a similar "hub" in a different neighborhood, etc. The global internet is just a (much) larger, more complex version (poluted by money and politics.) The job of an ISP is to forward traffic along a path towards the intended destination -- they do this using routing protocols (BGP) to know who's connected to who.
What component of the infrastructure design prevents people from bypassing the ISP in one big ad-hoc network?
The fact that, more often than not, the ISPs own the infrastructure. So, without running your own fiber or building your own microwave towers, you'll be dealing with them one way or another.
Utilities are regulated monopolies because of logisitical and physical limits. Imagine a world where you have 6 water companies all burying 16" mains in the right-of-way infront of your house. Or 12 power companies stringing lines all over the place. There's only so much room for water lines, gas lines, power lines, phone lines, sewers, etc. The more stuff hung from the poles or stuffed in the ground, the harder it is to keep track of it all to prevent them interfering with each other -- and eventually, something breaks and has to be repaired.
The internet is as much a commodity as gasoline. As NetZero says, they all take you to the same internet. What they leave out is how each gets you there. Bottom line, you have 3 ways of getting there... the phone wiring, the cable tv wiring, or an antenna of some kind. That limits you to DSL through the local telco -- or one of very few ISPs that maintain equipment in various COs, which is extremely rare these days -- a cable modem through the local cable company, or some radio based setup (cellular, WISP, HughsNet(tm), WiMax,...) that's generally very slow and expensive.
The choice of dozens of different dialup ISPs died over a decade ago. Dialup is wholesale these days -- just like DSL and cable. It's very uncommon for the ISP to actually own/operate the modem you're calling.
I have worked for and with many ISPs. I have never, EVER, seen a connection agreement that was billed by volume instead of speed. As someone else said, only consumers and stupid corps buy by the byte. And, for the record, I've never billed by the byte either... it's too much of a pain in the ass to track.
(There's no "reciprocal compensation" among ISPs like there is in the telco world.)
Actually, it's more like 1000:1. The issue isn't so much the transit points but the actual traffic on each QAM per node. Each channel supports ~40mbps (42.88) With 5M/384k there's enough download space for 8 people at full speed. It falls off gradually as more people are added, but everyone gets a pretty even cut of that 40M. As they ramp the speeds higher and higher, that means fewer and fewer ever get to see it. But it gets even worse. The higher your download rate, the higher your upload rate has to be to keep up with it. TCP requires ACKs of the data received or the transmitting stops. @5mbps, that's roughly 425 full size packets per second. With a 64k window, the remote side can transmit ~40 packets before stopping completely. So, in a perfect world -- and the internet is not such a place -- there must be at least 10 ACKs per second to keep the spice flowing; that's about 5kbps in ACKs only. Back in reality, it's about 5-10x that (ACK every 5 packets or so) or about 40kbps, minimum. (factor in queing and other traffic and that 40k starts looking more like 80k.) If your upstream path is saturated (or nearing it), your downstream throughput will start dropping substantially, because the stream of ACKs aren't getting through at the necessary rate.
Now here's the rub... they like to push download speed, but they don't pump the upload speeds much. Generally speaking, they are keeping the down:up at a functional point. However, the upstream capacity of the system is far less than the downstream -- 10mbps (1.0/1.1) or 30mbps (2.0). Unlike the equalizing effect with a staturated downstream channel -- due to queuing at the headend, when the upstream channel is full, everybody suffers. There's nothing the network can do about 75 cablemodems starved for airtime. 8M blistering downloads and 1K ssh sessions all suddenly slow to a crawl measured in single digit packets per second. It ain't pretty. (It's not 10base-2 bad, but it's something to avoid.)
No, everyone should be concerned about this. This is how we end up with the stories of people with $30,000 cellphone bills because they carried their iPhone onto a cruise ship (in port the entire time.) I want a flat-rate bill that isn't subject to insanity because someone hacked my laptop or tivo screws up and downloads a 2GB "tivocast" over and over. Bytes are not something people are generally aware they're using. Minutes on a cellphone... as you're holding the thing to your head, you should have some idea that you are burning minutes. Do you know how big this webpage is? How about that flash ad? Or the images? Or the RSS feeds being checked in the background? Or windows update fetching updates automatically... it's a very long list, and we are completely oblivious.
In most areas you can switch to earthlink using the same modem, same lines and switch the service over by phone in less than an hour.
... and STILL be a TWC customer. I *am* an Earthlink cablemodem subscriber -- have been, well, always -- and my bill comes from and payments go to... TIME WARNER. 100% of my $40/month goes to TIME WARNER. The only thing that makes it an Earthlink setup is the IP address I'm handed. (and all the "value added" crap I don't care about: web space, email accounts, usenet, etc.) When it breaks, I call TWC, not Earthlink.
If Verizon can do it and make money, TWC can as well.
TWC is already making ass bleeding amounts of money -- over $4 BILLION in profit last year. What they don't want to do is spend any money on operational upgrades to offer faster speeds -- a) it cuts into those massive profits, and b) it risks cutting down their profits on cable TV service. And as soon as TWC does this, every other ISP is going to see they can get away with it, too. And then *bam* the cost for broadband will go through the roof and no one will be able to aford to even read their email much less stream a 22kbps realmedia webcam video for 11secs.
Because it's hard to string cables to carry all of it. The rats nest of wiring in Japan and Korea would get people sued and executed here in the states. Heck, AT&T has been sued in many places for their Uverse cabnets -- aside from it being the size of a refridgerator (and white/beige), it's not that ugly.
Plus, there's all the red tape... monopolies, right-of-ways, agreements to use the utility poles, etc., etc. And there's a great deal of "not in my front yard" to fight as well.
Slashdot Mirror
You have an interesting definition of "not expensive". Fluke doesn't make anything that's cheap. And the NetTool is a $2500 device placing it well out of the budget of anyone who doesn't mess with cabling a lot.
Answer: Beldin (or whoever made the physical cable)
(I just crimped the ends on it.)
And I don't like the phrase "the cable went bad". Someone or something has to do something to it to make it stop working. A cable that's been plugged in and working for months or years will continue to do so right up to the point someone starts messing with it.
To that I say the people making your cables obviously aren't qualified to be making them.
Cables are either "good" or "bad"; they are either within spec or not. There's no such thing as a "marginal good" cable. I read your last sentence as meaning the cable is within spec as per whatever tested it, but then your network hardware isn't -- or your tester is crap.
Actually, it's a spec thing. The wiring specifications say 100m. So, by definition, any cable longer than 100m is out-of-spec. That said, in the old days, it read "200m node to node" which translates to 100m cables (100m node-switch x 2) And with certain switches (class I vs II) there could be an additional 2m or 5m between switches -- making the total 205m, max. (But that's from memory back when 100base-TX wasn't a standard.)
With the advent of FDX, line length has become less important. And people get away with cables far beyond spec. As long as there's enough signal at both ends.
I cannot speak to the cheap crap sold at Lowes, Home Depot, etc. However, the professional crimps use the same types of blades as the mass produced cables. Those "superior" factory made cables work loose over time, as well. And all those injection molded one's I've ever seen have much smaller actual crimps that make them much weaker. I've never had a cable I made break, but I've had numerous Belkin and Uniden factory made cables come completely apart and "work loose".
1 and 2 are made by humans. 3 is not. And it's very easy to tell them apart... the robots don't use the same crimps as they are injection molded. (and the wires are 100% straight and uniformly pushed to the back of the crimp.) The only way to make money making cables is to use robots. Lots of robots. People are far too expensive, slow, and prone to errors.
Right. The correct solution is to not aim a satelite transmitter through the edge of a building. The working solution is to sheild the T1 line and ground it sufficiently to keep out the RF causing errors. (which in this case was 3 places. admitedly, getting a proper ground on the 3rd floor of a building being baked by a dish is a problem in itself.)
But you go right a head and tell me how to fix problems you've never had to fix within systems where you have zero knowledge, experience, or expertise. "Well, that copper drain pipe should be grounding the flashing so there shouldn't be RF throughout the building."
That was not a simple "parabolic reflector". That was a transmitter for a radio station's uplink to it's transmitter station. So, far more power than most people are aware.
You can go on believing what ever you want. Until you see it for yourself, you are going to stand by what your books and theories tell you. It's people like you that make troubleshooting problems such a pain in the ass as you won't look for things you are certain are "impossible".
Well, it was cheaper than moving the building :-)
And no, putting a ground in one place will not "always" do it for you. Did you miss the part about a satelite transmitter next door? Aimed into the edge of the building? That's a lot of signal to keep out of the system... and it's literally everywhere.
Actually, I prefer my equipment not getting struck by lightning. As such, I use conduits were possible. And go out of my way to not hang stuff up in the air.
(And now that I think about it, all of the damage I've dealt with in the last few years has been at the hands of cable TV connections. 'tho several years ago, I had an ISDN NT-1 blown to bits, along with the cards in the RT and CO *grin* despite the entire 4.7 miles of cable being underground.)
Still wrong... OC3 is 155Mbps. [See also: The wiki]
Ah, I see we have another NCSU Alum from the Great Squirrel Power Outage of '90 (maybe 91?) "We know it was a squirrel due to the carbonized foot and tail left on the bus bar." Killed power to the entire main campus for most of the day.
Eventually, after you blow up enough switch ports
Or you learn to stop plugging lightning rods directly into your switch. Plug it into a patch panel / grounding block instead where it can be properly grounded vs. the tiny metal fingers found in RJ45 ports. (If it's direct strike, it's going to blow up whatever is connected at both ends no matter how well ground.)
... or tightly zip tie large bundles to trusses in the ceiling where you cannot see them or reach them without a crane. I've seen some zipped so tight it's cutting through the outer insulation.
And I can recall one location who's T1 was always flaky. It turned out to be the radio dish next door aimed across the top corner of the building where there was a copper drain pipe and copper flashing. It was enough RF noise to cause signalling issues. A liberal application of sheilded cables and ground straps fixed it up. (and no, there was nothing that could be done about the dish.)
(work in the telecom world long enough and you really can see everything :-))
The orginal article left out Cat3... coax (10b2) -> cat3 (10bT) -> cat5 (100bT) -> cat5e (1000bT) -> cat6 (10Gb, short runs) and cat6a (10Gb, full length) [and I'm leaving out 10b5/"thicknet"]
Your installation sounds like a very old phone wiring (cat3) that was repurposed for ethernet. Cat3 will work for 100bT, up to a point. It will not work very reliablly for 1000bT -- 'tho you may get lucky using cat3 patch cords. The 2 pair installations will not work beyond 100bT as all 4 pairs are used for GigE and 10Gb.
My nightmare is seeing one cable (usually cat3!) carrying *2* ethernets -- in clear disregard for documented cabling standards. And a coworker (network tech) who only crimped 2 pairs per end -- he'd leave the other 2 pairs out of the crimp on purpose... "you don't need those." (he had to carry around three cables... an ethernet cable, a T1 cable, and a serial cable because he wouldn't crimp one cable with all 4 pairs.)
I don't really have any idea what ISPs actually do or how internet infrastructure works.
Magic(tm).
It works just like your ad-hoc "connect to a friend" example. Your friend is then the ISP. He's responsible moving traffic to those who are directly connected to him. All the homes in your neighborhood connect to one house who connects to a similar "hub" in a different neighborhood, etc. The global internet is just a (much) larger, more complex version (poluted by money and politics.) The job of an ISP is to forward traffic along a path towards the intended destination -- they do this using routing protocols (BGP) to know who's connected to who.
What component of the infrastructure design prevents people from bypassing the ISP in one big ad-hoc network?
The fact that, more often than not, the ISPs own the infrastructure. So, without running your own fiber or building your own microwave towers, you'll be dealing with them one way or another.
Utilities are regulated monopolies because of logisitical and physical limits. Imagine a world where you have 6 water companies all burying 16" mains in the right-of-way infront of your house. Or 12 power companies stringing lines all over the place. There's only so much room for water lines, gas lines, power lines, phone lines, sewers, etc. The more stuff hung from the poles or stuffed in the ground, the harder it is to keep track of it all to prevent them interfering with each other -- and eventually, something breaks and has to be repaired.
The internet is as much a commodity as gasoline. As NetZero says, they all take you to the same internet. What they leave out is how each gets you there. Bottom line, you have 3 ways of getting there... the phone wiring, the cable tv wiring, or an antenna of some kind. That limits you to DSL through the local telco -- or one of very few ISPs that maintain equipment in various COs, which is extremely rare these days -- a cable modem through the local cable company, or some radio based setup (cellular, WISP, HughsNet(tm), WiMax, ...) that's generally very slow and expensive.
The choice of dozens of different dialup ISPs died over a decade ago. Dialup is wholesale these days -- just like DSL and cable. It's very uncommon for the ISP to actually own/operate the modem you're calling.
I have worked for and with many ISPs. I have never, EVER, seen a connection agreement that was billed by volume instead of speed. As someone else said, only consumers and stupid corps buy by the byte. And, for the record, I've never billed by the byte either... it's too much of a pain in the ass to track.
(There's no "reciprocal compensation" among ISPs like there is in the telco world.)
Actually, it's more like 1000:1. The issue isn't so much the transit points but the actual traffic on each QAM per node. Each channel supports ~40mbps (42.88) With 5M/384k there's enough download space for 8 people at full speed. It falls off gradually as more people are added, but everyone gets a pretty even cut of that 40M. As they ramp the speeds higher and higher, that means fewer and fewer ever get to see it. But it gets even worse. The higher your download rate, the higher your upload rate has to be to keep up with it. TCP requires ACKs of the data received or the transmitting stops. @5mbps, that's roughly 425 full size packets per second. With a 64k window, the remote side can transmit ~40 packets before stopping completely. So, in a perfect world -- and the internet is not such a place -- there must be at least 10 ACKs per second to keep the spice flowing; that's about 5kbps in ACKs only. Back in reality, it's about 5-10x that (ACK every 5 packets or so) or about 40kbps, minimum. (factor in queing and other traffic and that 40k starts looking more like 80k.) If your upstream path is saturated (or nearing it), your downstream throughput will start dropping substantially, because the stream of ACKs aren't getting through at the necessary rate.
Now here's the rub... they like to push download speed, but they don't pump the upload speeds much. Generally speaking, they are keeping the down:up at a functional point. However, the upstream capacity of the system is far less than the downstream -- 10mbps (1.0/1.1) or 30mbps (2.0). Unlike the equalizing effect with a staturated downstream channel -- due to queuing at the headend, when the upstream channel is full, everybody suffers. There's nothing the network can do about 75 cablemodems starved for airtime. 8M blistering downloads and 1K ssh sessions all suddenly slow to a crawl measured in single digit packets per second. It ain't pretty. (It's not 10base-2 bad, but it's something to avoid.)
No, everyone should be concerned about this. This is how we end up with the stories of people with $30,000 cellphone bills because they carried their iPhone onto a cruise ship (in port the entire time.) I want a flat-rate bill that isn't subject to insanity because someone hacked my laptop or tivo screws up and downloads a 2GB "tivocast" over and over. Bytes are not something people are generally aware they're using. Minutes on a cellphone... as you're holding the thing to your head, you should have some idea that you are burning minutes. Do you know how big this webpage is? How about that flash ad? Or the images? Or the RSS feeds being checked in the background? Or windows update fetching updates automatically... it's a very long list, and we are completely oblivious.
btw, uverse *is* DSL. It's fiber to the uverse node, but ADSL2+(?) to your house.
In most areas you can switch to earthlink using the same modem, same lines and switch the service over by phone in less than an hour.
... and STILL be a TWC customer. I *am* an Earthlink cablemodem subscriber -- have been, well, always -- and my bill comes from and payments go to... TIME WARNER. 100% of my $40/month goes to TIME WARNER. The only thing that makes it an Earthlink setup is the IP address I'm handed. (and all the "value added" crap I don't care about: web space, email accounts, usenet, etc.) When it breaks, I call TWC, not Earthlink.
If Verizon can do it and make money, TWC can as well.
TWC is already making ass bleeding amounts of money -- over $4 BILLION in profit last year. What they don't want to do is spend any money on operational upgrades to offer faster speeds -- a) it cuts into those massive profits, and b) it risks cutting down their profits on cable TV service. And as soon as TWC does this, every other ISP is going to see they can get away with it, too. And then *bam* the cost for broadband will go through the roof and no one will be able to aford to even read their email much less stream a 22kbps realmedia webcam video for 11secs.
Because it's hard to string cables to carry all of it. The rats nest of wiring in Japan and Korea would get people sued and executed here in the states. Heck, AT&T has been sued in many places for their Uverse cabnets -- aside from it being the size of a refridgerator (and white/beige), it's not that ugly.
Plus, there's all the red tape... monopolies, right-of-ways, agreements to use the utility poles, etc., etc. And there's a great deal of "not in my front yard" to fight as well.