I used to use RCN for my local service when I lived in a condo building they lit. I was very happy with their service once it was installed, and it was a LOT cheaper than any of the other alternatives.
I think you meant to say "packet switched" as opposed to "switched": the latter could also apply to circuit-switched TDM networks, which are engineered differently.
But seriously, think about it: a typical DSLAM is fed by an ATM DS3 (45m) or (rarely) OC3 (155m). How many 1.5m customers do you think can pull service from said uplink if all are trying to download at once? DOCSIS cable systems suffer from the same effect: exactly how much bandwidth do you think exists between Comcast and Google? The nature of packet switching is that communications are unscheduled, and tremendously vulnerable to short-lived congestion events. Well behaved protocols (read: http) back off when they encounter this. Badly behaved protocols (read: bittorrent) do not, and cause local congestion events to extend.
This is not new - what IS new is that ISPs engineered service under certain assumptions, and those facts on which those assumptions were based are changing.
Oversubscription is a very, very normal thing in service provider networks. Frame-Relay oversubscription is generally 15:1, ATM oversubscription was about 5:1, IP oversubscription is about 3:1. If you want truly non-oversubscribed bandwidth, prepare to pay a LOT more for it.
The problem isn't oversubscription, it's that the capacity management policies of some providers haven't caught up with the usage patterns of the customers. During peak periods, something's got to give.
Given that there are no providers selling truly non-oversubscribed bandwidth today, would you rather that the providers change their advertisements to say that, or raise their prices to sell dedicated bandwidth?
The "unintentional" killing you're referring to is closer to the American concept of "negligent homicide" - it was crimes for which the punishment was exile to once of the cities of refuge.
lo tirtzah in the Ten Commandments is a little bit broader than "don't murder," but it's a long distance away from "don't kill." Attempts to interpret it to mean "don't kill" are well-intentioned, but are not faithful to the actual meaning or historical interpretation of the text. If "don't kill" had been intended, lo ya'areg would have been used instead.
This is the standard Rabbinic interpretation of the phrase in the Ten Commandments - all of them refer to crimes for which one can receive the death penalty, so "don't steal" is limited to "don't kidnap [i.e. "don't steal people"]" - one cannot receive the death penalty for homicide which is not murder in Jewish law, so the that's another data point supporting the interpretation of the word as "murder."
I don't mind religion in my scifi any more than I mind chocolate in my peanut butter - what I do mind is garbled, incoherent religion. The problem with Moore's approach (the surprise reveal) is that the motivations for the characters get messed up on a re-watch. Consider instead the use of religion in Babylon 5: there are lots of deeply religious characters, and their religions are clearly different and important to them, but the religions are consistent from season to season.
And let's not forget that the central tenet of the series is that the Cylons were created by humans sometime in the not-so-distant past. Their religion should be no more shrouded in mystery to its adherents than, say, Mormonism or Christian Science (both about 150 years old).
I was a huge fan of the C64 version, and honestly, I think that better computer games are very rare - MULE was just brilliant, and far, far ahead of its time.
Apple just happened to get there first with the music and only-player-that-can-play-it combo (and even that shouldn't be patentable).
I'm mostly with you, although there are a few things about the iPod interface which strike me as patentable - the click-wheel, for instance. I don't know whether apple has a patent on it (and I'm waaaaay to lazy to look it up), but that is a very clever design which I hadn't seen before in handheld devices. But of course, the click-wheel is an object, which immediately renders it an appropriate subject of a patent.
but is it enough to datamine substantially all call records? No.
Here's why: providing a bulk CDR (call detail record) is an SS7 function. There isn't anything you can do from your handset to query the system to provide you with this information. The closest you can get is the *69-style commands, but those are not appropriate for bulk data. If a LEA wanted to acquire CDR information, the easiest way to do it would be to ask the telcos, rather than try to acquire it via a live feed. If they were datamining as you suggest, they'd want a circuit not into the switching system, but rather into the accounting system...
The most likely explanation for DS3s or other low-speed (yes, a DS3 is low-speed to a serious carrier) circuits between the FBI and a carrier is that it would be used for the specific monitoring of individuals after permission is granted. FBI goes to telco, and says "monitor this line." Telco says "what should we do with the data?" FBI says, "send it down our DS3." Telco either says "ok" or "show us the warrant."
What you're worried about is the strength of the "permission is granted" part, rather than the technical details. That's necessarily a matter of policy, and I agree that Congress has done a crummy job of creating clear regulations on the matter. Lots of hot air, yes, but little clarity.
For reference purposes, the AT&T, Sprint, and Verizon network backbones use NxOC-192 (10Gbps) and NxOC-768 (40Gbps) SONET circuits. Of course, that includes both voice AND data, but it should show the general irrelevance of a single DS3.
I've never seen an OC-24: the more common value in the US is an OC-48 (2.4Gbps). A good rule of thumb for getting the relative size of these pipes is that the number after the OC- represents roughly the number of DS3s which can be carried on the optical path. Of course, the DS3 is encapsulated as an STS-1 channel (53Mbps) on the larger circuit...
I've encountered a related problem in my (non-programming) line of work. I'm in networking, and I've probably given 100 technical interviews over the past 2 years. Of those, I've only recommended hiring 4 or 5 of the candidates. Why not? Because when someone puts "OSPF" on their resume, I expect them to be able to intelligently talk about its utility and shortcomings, and compare it with other IGPs.
A bunch of the candidates I've spoken to have been egregious fakers, and it's really, really hard to find the good ones among the not-so-good.
Roads have existed long before cars and used by horse, carriage, and other methods of transportation for millenia. Yet, only recently have we tried this mantra of "it's not a right, it's a privilege".
Go further back than all that: it used to be that roads were built by individual rulers or individual societies, and there were LOTS of toll roads - the tolls did things like finance the soldiers who would protect them. The word you're looking for is path not road.
In the USA, roads are built by the state or private individuals, and the state or private individuals are allowed to place conditions on the public use of those roads. Typical conditions would be "use a liscenced vehicle" or "pay a toll." If one is unwilling to comply with those conditions, and attempt to use the road anyway, then one can be hauled off to jail. Sounds like it's not quite as inalienable right as you implied...
The "green" block in the upper right is 240/4 "class E" space, which was and is IANA-reserved for "experiments." The problem is that most computers currently in use cannot either accept one of these addresses or route to it. Go ahead - try, and see how it works. (NANOG had a long thread about reclaiming that space, and the net of it was that it wasn't worth the effort).
Public IP space is not only important for public-facing networks - it's when it absolutely needs to be globally unique: think network management, or when you acquire a bunch of other companies - 3 and 4 layer NAT really sucks.
IBM also uses those addresses for their partners, of which there are zillions. They may not need a/8, but they probably need at least a/9, which means that any actual renumbering and give-back wouldn't net very much
AT&T, definitely (how many customers do you think AT&T has?)
DEC -> Compaq -> HP... possibly - they do a LOT of stuff
MIT? Yes.
IBM? Also yes.
So that leaves Apple, Ford and Xerox as possibilities.
The current burn rate chews through about/6 per year, so even if everyone in that list besides AT&T and IBM (who are in fact ISPs) gave back theirs, that buys about an additional year.
The presentation was at NANOG, which just concluded on Wednesday of this week.
If all of the class-A space allocated to companies was revoked (there are actually some companies which really DO use their whole class-A, and the legality of revoking it would be pretty questionable (they don't have a contractual relationsihp with ARIN, for instance), that would buy about another year given the current burn rate. No matter when it runs out, it's running out, and we should be prepared for it.
yeah, it was pretty ghetto. I was very excited to upgrade to 9600 so that Mosaic would work. On the flip side, tcsh/pine weren't so bad over slow connections...
Hah! Pretty close - we didn't carry on a torrid online romance [2400 baud romance = teh sux0r], we just arranged to meet (in a "meet new friends" sort of way - totally platonic), and once I saw her, I fell head-over-heels. She thought I was okay, but a few years of dogged pursuit won her heart. That and a mutual friend talking me up to her, of course (I owe that friend BIG)...
Slashdot Mirror
I used to use RCN for my local service when I lived in a condo building they lit. I was very happy with their service once it was installed, and it was a LOT cheaper than any of the other alternatives.
Voting with your feet is tough in a lot of places - there are a very small number of actual service providers to choose from.
Theodore Sturgeon got very into the difference between Ethics and Morality in More than Human, and had some interesting things to say about it.
I think you meant to say "packet switched" as opposed to "switched": the latter could also apply to circuit-switched TDM networks, which are engineered differently.
But seriously, think about it: a typical DSLAM is fed by an ATM DS3 (45m) or (rarely) OC3 (155m). How many 1.5m customers do you think can pull service from said uplink if all are trying to download at once? DOCSIS cable systems suffer from the same effect: exactly how much bandwidth do you think exists between Comcast and Google? The nature of packet switching is that communications are unscheduled, and tremendously vulnerable to short-lived congestion events. Well behaved protocols (read: http) back off when they encounter this. Badly behaved protocols (read: bittorrent) do not, and cause local congestion events to extend.
This is not new - what IS new is that ISPs engineered service under certain assumptions, and those facts on which those assumptions were based are changing.
Oversubscription is a very, very normal thing in service provider networks. Frame-Relay oversubscription is generally 15:1, ATM oversubscription was about 5:1, IP oversubscription is about 3:1. If you want truly non-oversubscribed bandwidth, prepare to pay a LOT more for it.
The problem isn't oversubscription, it's that the capacity management policies of some providers haven't caught up with the usage patterns of the customers. During peak periods, something's got to give.
Given that there are no providers selling truly non-oversubscribed bandwidth today, would you rather that the providers change their advertisements to say that, or raise their prices to sell dedicated bandwidth?
The "unintentional" killing you're referring to is closer to the American concept of "negligent homicide" - it was crimes for which the punishment was exile to once of the cities of refuge.
lo tirtzah in the Ten Commandments is a little bit broader than "don't murder," but it's a long distance away from "don't kill." Attempts to interpret it to mean "don't kill" are well-intentioned, but are not faithful to the actual meaning or historical interpretation of the text. If "don't kill" had been intended, lo ya'areg would have been used instead.
This is the standard Rabbinic interpretation of the phrase in the Ten Commandments - all of them refer to crimes for which one can receive the death penalty, so "don't steal" is limited to "don't kidnap [i.e. "don't steal people"]" - one cannot receive the death penalty for homicide which is not murder in Jewish law, so the that's another data point supporting the interpretation of the word as "murder."
gawk|uncompress|unzip|nice|head|strip|touch|finger|mount|fsck|more|yes|gasp|umount|sleep
I don't mind religion in my scifi any more than I mind chocolate in my peanut butter - what I do mind is garbled, incoherent religion. The problem with Moore's approach (the surprise reveal) is that the motivations for the characters get messed up on a re-watch. Consider instead the use of religion in Babylon 5: there are lots of deeply religious characters, and their religions are clearly different and important to them, but the religions are consistent from season to season.
And let's not forget that the central tenet of the series is that the Cylons were created by humans sometime in the not-so-distant past. Their religion should be no more shrouded in mystery to its adherents than, say, Mormonism or Christian Science (both about 150 years old).
Word.
I was a huge fan of the C64 version, and honestly, I think that better computer games are very rare - MULE was just brilliant, and far, far ahead of its time.
Heck, for that matter, Frank Zappa talked about an almost identical idea in the interviews that became The Real Frank Zappa Book, published in 1990.
I'm mostly with you, although there are a few things about the iPod interface which strike me as patentable - the click-wheel, for instance. I don't know whether apple has a patent on it (and I'm waaaaay to lazy to look it up), but that is a very clever design which I hadn't seen before in handheld devices. But of course, the click-wheel is an object, which immediately renders it an appropriate subject of a patent.
Here's why: providing a bulk CDR (call detail record) is an SS7 function. There isn't anything you can do from your handset to query the system to provide you with this information. The closest you can get is the *69-style commands, but those are not appropriate for bulk data. If a LEA wanted to acquire CDR information, the easiest way to do it would be to ask the telcos, rather than try to acquire it via a live feed. If they were datamining as you suggest, they'd want a circuit not into the switching system, but rather into the accounting system...
The most likely explanation for DS3s or other low-speed (yes, a DS3 is low-speed to a serious carrier) circuits between the FBI and a carrier is that it would be used for the specific monitoring of individuals after permission is granted. FBI goes to telco, and says "monitor this line." Telco says "what should we do with the data?" FBI says, "send it down our DS3." Telco either says "ok" or "show us the warrant."
What you're worried about is the strength of the "permission is granted" part, rather than the technical details. That's necessarily a matter of policy, and I agree that Congress has done a crummy job of creating clear regulations on the matter. Lots of hot air, yes, but little clarity.
For reference purposes, the AT&T, Sprint, and Verizon network backbones use NxOC-192 (10Gbps) and NxOC-768 (40Gbps) SONET circuits. Of course, that includes both voice AND data, but it should show the general irrelevance of a single DS3.
I've never seen an OC-24: the more common value in the US is an OC-48 (2.4Gbps). A good rule of thumb for getting the relative size of these pipes is that the number after the OC- represents roughly the number of DS3s which can be carried on the optical path. Of course, the DS3 is encapsulated as an STS-1 channel (53Mbps) on the larger circuit...
A GSM half-rate channel is 5.6Kbps (a fullrate channel is twice that, but let's look at the most extreme case). A DS3 = 45 Mbps. 45Mbps = 45000Kbps
45000Kbps / 5.6Kbps = 8037 simultaneous calls supported on a DS3, assuming 0% overhead, protocol, encryption, and that all calls are half-rate.
VZW and ATTW have subscriber counts in the millions.
Whatever the legality or circumstance of this, a single DS3 is hardly wholesale snooping.
I've encountered a related problem in my (non-programming) line of work. I'm in networking, and I've probably given 100 technical interviews over the past 2 years. Of those, I've only recommended hiring 4 or 5 of the candidates. Why not? Because when someone puts "OSPF" on their resume, I expect them to be able to intelligently talk about its utility and shortcomings, and compare it with other IGPs.
A bunch of the candidates I've spoken to have been egregious fakers, and it's really, really hard to find the good ones among the not-so-good.
Roads have existed long before cars and used by horse, carriage, and other methods of transportation for millenia. Yet, only recently have we tried this mantra of "it's not a right, it's a privilege".
Go further back than all that: it used to be that roads were built by individual rulers or individual societies, and there were LOTS of toll roads - the tolls did things like finance the soldiers who would protect them. The word you're looking for is path not road.
In the USA, roads are built by the state or private individuals, and the state or private individuals are allowed to place conditions on the public use of those roads. Typical conditions would be "use a liscenced vehicle" or "pay a toll." If one is unwilling to comply with those conditions, and attempt to use the road anyway, then one can be hauled off to jail. Sounds like it's not quite as inalienable right as you implied...
The "green" block in the upper right is 240/4 "class E" space, which was and is IANA-reserved for "experiments." The problem is that most computers currently in use cannot either accept one of these addresses or route to it. Go ahead - try, and see how it works. (NANOG had a long thread about reclaiming that space, and the net of it was that it wasn't worth the effort).
Public IP space is not only important for public-facing networks - it's when it absolutely needs to be globally unique: think network management, or when you acquire a bunch of other companies - 3 and 4 layer NAT really sucks.
IBM also uses those addresses for their partners, of which there are zillions. They may not need a /8, but they probably need at least a /9, which means that any actual renumbering and give-back wouldn't net very much
AT&T, definitely (how many customers do you think AT&T has?)
/6 per year, so even if everyone in that list besides AT&T and IBM (who are in fact ISPs) gave back theirs, that buys about an additional year.
DEC -> Compaq -> HP... possibly - they do a LOT of stuff
MIT? Yes.
IBM? Also yes.
So that leaves Apple, Ford and Xerox as possibilities.
The current burn rate chews through about
Not enough to make a meaningful difference...
Lots of folks do this: google "virtual hosting" - it's been around since the mid-90s.
However, it doesn't solve the depletion problem, becuase most of the depletion comes from access users not content providers.
Do you have any idea how big either one of them are? They do actually USE those addresses, you know.
Also, that doesn't buy much time - it's a lot of work for very little benefit.
The presentation was at NANOG, which just concluded on Wednesday of this week.
If all of the class-A space allocated to companies was revoked (there are actually some companies which really DO use their whole class-A, and the legality of revoking it would be pretty questionable (they don't have a contractual relationsihp with ARIN, for instance), that would buy about another year given the current burn rate. No matter when it runs out, it's running out, and we should be prepared for it.
-David
yeah, it was pretty ghetto. I was very excited to upgrade to 9600 so that Mosaic would work. On the flip side, tcsh/pine weren't so bad over slow connections...
Hah! Pretty close - we didn't carry on a torrid online romance [2400 baud romance = teh sux0r], we just arranged to meet (in a "meet new friends" sort of way - totally platonic), and once I saw her, I fell head-over-heels. She thought I was okay, but a few years of dogged pursuit won her heart. That and a mutual friend talking me up to her, of course (I owe that friend BIG)...
not anymore there aren't...