1 Mbps =~= 10 GB/day = 300GB/month. If your cable company puts a bandwidth cap on your service, it's effectively slowing you down to less than DSL speeds.
Blocking inbound SMTP isn't going to prevent any spam; it's just going to force people to use commercial email services to get their mail. No excuse for it.
There are three kinds of users who send outbound SMTP
Legitimate home email users.
Infected zombies sending spam.
Spammers using home systems.
Many ISPs have a policy of "block SMTP by default, but allow it if the user requests", which keeps out the zombies. It does force them to deal with occasional spam complaints because of customers who spam on purpose, but they're blockable.
I don't use Netflix, but here in Silicon Valley my 3 Mbps DSL is perfectly capable of playing standardish-definition TV from TV network websites, as well as playing YouTube. If I were a sports fan I might care about getting HDTV sports over the net instead of cable TV, since I assume Comcast's sports channel selections are as lame as their non-sports TV channel selections and sports actually does benefit from the higher resolution.
Back in the 90s and early 2000s, we were trying to sell businesses on using 8kbps G.729 calls from IP PBXs instead of 64kbps telco voice, and they would whine about Mean Opinion Scores and latency (and didn't get that India just wasn't going to get any closer and the speed of light wasn't going to change.) Cell phones convinced most of those people that they didn't really need to care - GSM was 13kbps or 6.5, and your office PBX phones had much better microphones than a typical cellphone and usually didn't have wind noise and trucks going by in the background.
I did have one friend who kept an analog cell phone around for a long time after most people had switched over to digital, because he spent a lot of time out in the mountains and back-country where there wasn't yet much cellphone signal, and a bad analog call was noisy, while a bad digital call just wouldn't stay connected.
When you make a VOIP connection, you're signalling to the network that you want to do that, it finds you the IP address and port number, either for a gateway into the old telco network or else for the phone you're calling. That's not getting you out to the public internet, though if you've got another friend with another rooted phone who's also got an active wifi connection, maybe you could do something useful with it.
But remember the other signalling that's going on, between your phone and the cell tower, which is keeping track of how much bandwidth you're sending - you'll have to make it think you're on a voice call also, if the voice prices are managed separately from the data prices.
Jitterbug's been a great phone for my mom. Her vision's not very good, so she doesn't bother texting (she'd need to hold a magnifying glass in one hand and use the phone with the other) , and she's stubborn enough she doesn't like to carry the phone around unless she expects to need it (e.g. going somewhere that she'll need to call a taxi), but it's reliable, does voice just fine, has big buttons for dialing, and makes free long-distance calls (so she doesn't bother buying long-distance from her landline telco any more.) The only way a smartphone would do her any good would be if Siri or equivalent could do everything, not just almost everything.
Vanilla telco G.711 is 64kbps, and it's what the digital parts of a telco voice call use. G.729 codecs, mostly used for PBXs, use 8 kbps. GSM had several codecs, including 13.3, 6.5, and less. The problem with all of these is that they need to send lots of packets per second to minimize latency - typically 20 or 30 - so the transport protocol overhead is usually several times higher than the actual voice payload, between IP, UDP, RDP, plus any layer-2 overhead (Ethernet's huge, or ATM's a bit less, if your DSL is using ATM.) In some limited environments there are ways around it, e.g. using CSLIP to do header compression on modems or whatever, but I don't know if most of the LTE carriers are going to do anything like that, or if they're just going to run native VOIP.
The overhead gets a lot worse if you need crypto and implement it naively using IPSEC, because you end up with a couple more layers of headers on top. It would obviously be better to have an encrypted payload standard, because then all it would cost you is some setup key exchange, but the telcos haven't had a big reason to do that, and most of the IP PBX vendors haven't either.
My GPS uses 2G cellular to get its traffic data and gas prices and to do Google search for destinations. It's going away next year, because Garmin's contract with the cellphone carrier isn't going to be renewed:-(.
Carriers really want that spectrum back, and 3G and LTE are much more efficient in terms of data bitrate per MHz of radio, plus they want to cut down on the number of separate types of equipment (not only for equipment costs, but also because keeping two separate channels of data is much less efficient than one fat channel.)
I don't know the VoLTE protocols, but for regular PBX-style VOIP, the voice compression is good enough and the voice payload in the packets is small enough that most of the bandwidth is used for IP/UDP/RDP headers, not the actual voice. There are way too many standards to choose from, but most of them run about 5KB/sec or less (that's bytes, not bits), so about 300 KB/min, or about 3000 min for 1 GB. There are people who use that much voice time, but not many:-) I'd expect that for a while you'll see multiple different standards for handling hd-mobile-to-hd-mobile, sd-mobile-to-sd-mobile, mobile-to-wireline, mobile-to-other-mobile-carrier, etc.
Back around 1990, I went to a technology talk by a guy from MCI who thought that the conflicting economics of offering voice and video on the same network were going to be a serious problem for telcos - video at the time meant ~1.5-3 Mbps corporate teleconferencing, and either you could price video too high to sell much of it, or you could sell T1 bandwidth cheaply enough to make videoconferencing affordable, in which case you'd undercut your voice pricing because companies would buy your video T1s to interconnect their PBXs for cheap. Better video compression got us out of that hole for a few years (384kbps or especially 128kbps video didn't cause that much trouble), but the Internet came along and started doing the same technological undercutting, VOIP started becoming feasible, etc. Mobile phones gave us a way to charge lots of money per minute again, but Moore's Law is still relentless.
Disclaimer: I do work for AT&T, but I do computer security, not mobile phones, so I have no idea what they're planning to charge for this, this is my own opinion, not the company's, blah blah blah. On the other hand, I have been doing various kinds of telco things for many generations of technology:-)
While the Reagan Administration really wanted to get most of Corporate America doing that, as a tool in the War On Drugs, it was hardly universal, partly to allow companies to go way overboard without the government having to take responsibility.
Cygnus Solutions, a company that did open-source gcc and other GNU work, had a contract supporting the state of California with compilers, so they were required to have a corporate drug policy and have it posted up on the same board as the minimum wage notices, etc. There was no requirement for the policy to be anything specific, including testing, and the company eventually decided on an official policy that if you bring illegal drugs to the workplace, you have to offer to share them with your coworkers, and posted it. I'm not sure how often the policy was actually followed, but I know some obvious people to ask:-)
Back when I was reading the Internet on a 14.4-kbps modem, the bandwidth used by ad banners was annoying, but you could block some of them with a hosts file, and the others weren't really that annoying unless they were using blink tags or animated GIFs. (Popups were annoying enough that most people blocked them pretty quickly.)
But sorry, if my browser is going to run random Javascript or Flash, it means my browser is going to run slowly and unreliably, and there's a risk of malicious content, and it's not safe to allow that kind of stuff.
As somebody else pointed out, either you or the person you were talking to was thinking of Merkle-Hellman, not Diffie-Hellman.
NP-hard problems are attractive, because they take polynomial time to verify if you know the right piece of data (i.e. the key) and exponentialish time if you don't. But it turns out that there are a lot of them that either can't be usefully turned into an encryption algorithm, or usually aren't more than polynomially hard if you do, because the version of the problem that got you public-key encryption is either some subset problem that isn't exponentially hard, or is only hard sometimes but not always, or else because turning the problem from a decision system into an encryption system took you exponential amounts of work (i.e. wasn't useful.)
So far Discrete Logarithm (over various groups, including modulo-prime arithmetic and elliptic curves) and factoring have been the most useful problems, giving us Diffie-Hellman and RSA and a few signature systems.
It's a way to model attacks by [current favorite threat] or [some other threat] without the risk that you'll get caught yet again using training material that's racist or religiously prejudiced or stupidly outdated, avoids the political problems of using training material where the "enemy" is now one of our allies (like the Germans or Russians or in some decades, Iraqis), and eliminates the problem that the training-material enemy is some national or ethnic group that some of your soldiers happen to belong to. And it means you don't have to do sensitivity training for the people who write your training material.
My nephew spent a number of his teenage years on the couch in front of the TV and various game consoles, earphones in for talking to his buddies, shooting zombies or enemy soldiers or each other or whoever they felt like shooting that afternoon.
Well, duh. Or maybe San Antonio. Of course, either way, once you walk outside you're in Texas weather, not that Sacramento's much better. But there's certainly no reason to want to be in Dallas or Houston; it's like LA without the culture or weather.
Yeah, the valley isn't the city. But here in Mountain View, there are usually about 25 cuisines of restaurants on our 4 blocks of downtown restaurant zone, and you can find a few more in Palo Alto or Sunnyvale, plus a lot more range of Indian and Korean farther down El Camino. We don't have much in the way of nightclubs, but there's plenty of choices of music jams around. I do have one friend who was living in San Jose and decided there wasn't enough social life down there (i.e. chances to meet women), so he moved up to the city and found that the women in the bars in his new neighborhood were also there to meet women, but eventually got to know somebody from his musician circles.
Phoenix area has a lot of data centers, which got built there because there's no risk of earthquakes, floods, or hurricanes, and at least used to have a lot of chip factories because it was cheap and had minimal environmental regulation. Not sure how much that's still the case; if I were going to move to Arizona, I'd much prefer Tucson, which is relatively civilized.
There are a lot of people in the Bay Area who already have commutes that hopelessly suck. Sacramento's just a bit farther away from San Francisco than places like Brentwood are from San Jose - it's 90 miles, which Google Maps says is about 1.5 hours in current traffic (though about 5 hours at rush hour.) And look at the surrounding communities - Roseville (big HP campus there and SF banks), Folsom (Intel), Rancho Cordoba (insurance and health care companies along freeway), Elk Grove (Apple), and bunches of other Silicon Valley companies that have large branch offices because it was close enough to Silicon Valley and the land was cheap enough to build data centers.
There have been some cultural changes out in that area as well since the time I was visiting occasional customers out there. Until Starbucks got to town, there was a local conspiracy not to sell any coffee strong enough to wake up a state bureaucrat. Other than one Lebanese restaurant, you simply couldn't get espresso, and the coffee at state office buildings was watery swill that's about like what McDonald's gets while they're washing their coffee pots. The stuff at gas stations near the freeway wasn't thick enough to burn.
It's not uncommon for some kinds of startups to move to the Lake Tahoe area when they're about to make some money, so that they get Nevada's near-zero taxes instead of getting hit with California taxes. You can still drive down to civilization if you need to see people, and if you were originally Easterners instead of native Californians the idea of snow isn't scary.
Remember when there was computer hardware? Companies like Sun and Silicon Graphics and a bunch of little Motorola 680x0 workstation companies?
Yeah, that boom had ended when I moved here in the early 90s, but there was still enough interesting culture and good weather to justify moving out from the east coast, even though the Internet meant you really could work from anywhere in the world you wanted. I caught the tail end of the housing slump (which meant my house in NJ made a good down-payment on a condo out here.)
Those didn't exist at my first few jobs; smokers could still smoke in the office back then:-(. But the one place I worked in the early 90s which had smokers, I'd usually join them outside for a smoke break and just stand upwind, because the break-and-socialization was an important part of the job.
Believe me, I have plenty of experience with freezing rain and cars; I currently deal with the problem by living in California.:-)
And yeah, I do prefer simple non-electronic keys - my previous two cars were Chevy vans, which meant that not only was it easy to replace keys, and carry a AAA-made plastic key in my wallet in case I got locked out, but I could also break in with a screwdriver without much trouble.
I've been annoyed that I haven't been able to see the original movie since it was first in theaters back in the 70s; SW4:ANH just isn't the same thing.
Slashdot Mirror
1 Mbps =~= 10 GB/day = 300GB/month. If your cable company puts a bandwidth cap on your service, it's effectively slowing you down to less than DSL speeds.
Blocking inbound SMTP isn't going to prevent any spam; it's just going to force people to use commercial email services to get their mail. No excuse for it.
There are three kinds of users who send outbound SMTP
Many ISPs have a policy of "block SMTP by default, but allow it if the user requests", which keeps out the zombies. It does force them to deal with occasional spam complaints because of customers who spam on purpose, but they're blockable.
I don't use Netflix, but here in Silicon Valley my 3 Mbps DSL is perfectly capable of playing standardish-definition TV from TV network websites, as well as playing YouTube. If I were a sports fan I might care about getting HDTV sports over the net instead of cable TV, since I assume Comcast's sports channel selections are as lame as their non-sports TV channel selections and sports actually does benefit from the higher resolution.
Back in the 90s and early 2000s, we were trying to sell businesses on using 8kbps G.729 calls from IP PBXs instead of 64kbps telco voice, and they would whine about Mean Opinion Scores and latency (and didn't get that India just wasn't going to get any closer and the speed of light wasn't going to change.) Cell phones convinced most of those people that they didn't really need to care - GSM was 13kbps or 6.5, and your office PBX phones had much better microphones than a typical cellphone and usually didn't have wind noise and trucks going by in the background.
I did have one friend who kept an analog cell phone around for a long time after most people had switched over to digital, because he spent a lot of time out in the mountains and back-country where there wasn't yet much cellphone signal, and a bad analog call was noisy, while a bad digital call just wouldn't stay connected.
When you make a VOIP connection, you're signalling to the network that you want to do that, it finds you the IP address and port number, either for a gateway into the old telco network or else for the phone you're calling. That's not getting you out to the public internet, though if you've got another friend with another rooted phone who's also got an active wifi connection, maybe you could do something useful with it.
But remember the other signalling that's going on, between your phone and the cell tower, which is keeping track of how much bandwidth you're sending - you'll have to make it think you're on a voice call also, if the voice prices are managed separately from the data prices.
Jitterbug's been a great phone for my mom. Her vision's not very good, so she doesn't bother texting (she'd need to hold a magnifying glass in one hand and use the phone with the other) , and she's stubborn enough she doesn't like to carry the phone around unless she expects to need it (e.g. going somewhere that she'll need to call a taxi), but it's reliable, does voice just fine, has big buttons for dialing, and makes free long-distance calls (so she doesn't bother buying long-distance from her landline telco any more.) The only way a smartphone would do her any good would be if Siri or equivalent could do everything, not just almost everything.
Vanilla telco G.711 is 64kbps, and it's what the digital parts of a telco voice call use. G.729 codecs, mostly used for PBXs, use 8 kbps. GSM had several codecs, including 13.3, 6.5, and less. The problem with all of these is that they need to send lots of packets per second to minimize latency - typically 20 or 30 - so the transport protocol overhead is usually several times higher than the actual voice payload, between IP, UDP, RDP, plus any layer-2 overhead (Ethernet's huge, or ATM's a bit less, if your DSL is using ATM.) In some limited environments there are ways around it, e.g. using CSLIP to do header compression on modems or whatever, but I don't know if most of the LTE carriers are going to do anything like that, or if they're just going to run native VOIP.
The overhead gets a lot worse if you need crypto and implement it naively using IPSEC, because you end up with a couple more layers of headers on top. It would obviously be better to have an encrypted payload standard, because then all it would cost you is some setup key exchange, but the telcos haven't had a big reason to do that, and most of the IP PBX vendors haven't either.
My GPS uses 2G cellular to get its traffic data and gas prices and to do Google search for destinations. It's going away next year, because Garmin's contract with the cellphone carrier isn't going to be renewed :-(.
Carriers really want that spectrum back, and 3G and LTE are much more efficient in terms of data bitrate per MHz of radio, plus they want to cut down on the number of separate types of equipment (not only for equipment costs, but also because keeping two separate channels of data is much less efficient than one fat channel.)
I don't know the VoLTE protocols, but for regular PBX-style VOIP, the voice compression is good enough and the voice payload in the packets is small enough that most of the bandwidth is used for IP/UDP/RDP headers, not the actual voice. There are way too many standards to choose from, but most of them run about 5KB/sec or less (that's bytes, not bits), so about 300 KB/min, or about 3000 min for 1 GB. There are people who use that much voice time, but not many :-) I'd expect that for a while you'll see multiple different standards for handling hd-mobile-to-hd-mobile, sd-mobile-to-sd-mobile, mobile-to-wireline, mobile-to-other-mobile-carrier, etc.
Back around 1990, I went to a technology talk by a guy from MCI who thought that the conflicting economics of offering voice and video on the same network were going to be a serious problem for telcos - video at the time meant ~1.5-3 Mbps corporate teleconferencing, and either you could price video too high to sell much of it, or you could sell T1 bandwidth cheaply enough to make videoconferencing affordable, in which case you'd undercut your voice pricing because companies would buy your video T1s to interconnect their PBXs for cheap. Better video compression got us out of that hole for a few years (384kbps or especially 128kbps video didn't cause that much trouble), but the Internet came along and started doing the same technological undercutting, VOIP started becoming feasible, etc. Mobile phones gave us a way to charge lots of money per minute again, but Moore's Law is still relentless.
Disclaimer: I do work for AT&T, but I do computer security, not mobile phones, so I have no idea what they're planning to charge for this, this is my own opinion, not the company's, blah blah blah. On the other hand, I have been doing various kinds of telco things for many generations of technology :-)
While the Reagan Administration really wanted to get most of Corporate America doing that, as a tool in the War On Drugs, it was hardly universal, partly to allow companies to go way overboard without the government having to take responsibility.
Cygnus Solutions, a company that did open-source gcc and other GNU work, had a contract supporting the state of California with compilers, so they were required to have a corporate drug policy and have it posted up on the same board as the minimum wage notices, etc. There was no requirement for the policy to be anything specific, including testing, and the company eventually decided on an official policy that if you bring illegal drugs to the workplace, you have to offer to share them with your coworkers, and posted it. I'm not sure how often the policy was actually followed, but I know some obvious people to ask :-)
Back when I was reading the Internet on a 14.4-kbps modem, the bandwidth used by ad banners was annoying, but you could block some of them with a hosts file, and the others weren't really that annoying unless they were using blink tags or animated GIFs. (Popups were annoying enough that most people blocked them pretty quickly.)
But sorry, if my browser is going to run random Javascript or Flash, it means my browser is going to run slowly and unreliably, and there's a risk of malicious content, and it's not safe to allow that kind of stuff.
I did misread it as "NSA's Plan" rather than "NASA's Plan" the first time.
As somebody else pointed out, either you or the person you were talking to was thinking of Merkle-Hellman, not Diffie-Hellman.
NP-hard problems are attractive, because they take polynomial time to verify if you know the right piece of data (i.e. the key) and exponentialish time if you don't. But it turns out that there are a lot of them that either can't be usefully turned into an encryption algorithm, or usually aren't more than polynomially hard if you do, because the version of the problem that got you public-key encryption is either some subset problem that isn't exponentially hard, or is only hard sometimes but not always, or else because turning the problem from a decision system into an encryption system took you exponential amounts of work (i.e. wasn't useful.)
So far Discrete Logarithm (over various groups, including modulo-prime arithmetic and elliptic curves) and factoring have been the most useful problems, giving us Diffie-Hellman and RSA and a few signature systems.
It's a way to model attacks by [current favorite threat] or [some other threat] without the risk that you'll get caught yet again using training material that's racist or religiously prejudiced or stupidly outdated, avoids the political problems of using training material where the "enemy" is now one of our allies (like the Germans or Russians or in some decades, Iraqis), and eliminates the problem that the training-material enemy is some national or ethnic group that some of your soldiers happen to belong to. And it means you don't have to do sensitivity training for the people who write your training material.
My nephew spent a number of his teenage years on the couch in front of the TV and various game consoles, earphones in for talking to his buddies, shooting zombies or enemy soldiers or each other or whoever they felt like shooting that afternoon.
The theater quality of the film version I saw back in the 1970s was just fine.
Well, duh. Or maybe San Antonio. Of course, either way, once you walk outside you're in Texas weather, not that Sacramento's much better. But there's certainly no reason to want to be in Dallas or Houston; it's like LA without the culture or weather.
Yeah, the valley isn't the city. But here in Mountain View, there are usually about 25 cuisines of restaurants on our 4 blocks of downtown restaurant zone, and you can find a few more in Palo Alto or Sunnyvale, plus a lot more range of Indian and Korean farther down El Camino. We don't have much in the way of nightclubs, but there's plenty of choices of music jams around. I do have one friend who was living in San Jose and decided there wasn't enough social life down there (i.e. chances to meet women), so he moved up to the city and found that the women in the bars in his new neighborhood were also there to meet women, but eventually got to know somebody from his musician circles.
Phoenix area has a lot of data centers, which got built there because there's no risk of earthquakes, floods, or hurricanes, and at least used to have a lot of chip factories because it was cheap and had minimal environmental regulation. Not sure how much that's still the case; if I were going to move to Arizona, I'd much prefer Tucson, which is relatively civilized.
There are a lot of people in the Bay Area who already have commutes that hopelessly suck. Sacramento's just a bit farther away from San Francisco than places like Brentwood are from San Jose - it's 90 miles, which Google Maps says is about 1.5 hours in current traffic (though about 5 hours at rush hour.) And look at the surrounding communities - Roseville (big HP campus there and SF banks), Folsom (Intel), Rancho Cordoba (insurance and health care companies along freeway), Elk Grove (Apple), and bunches of other Silicon Valley companies that have large branch offices because it was close enough to Silicon Valley and the land was cheap enough to build data centers.
There have been some cultural changes out in that area as well since the time I was visiting occasional customers out there. Until Starbucks got to town, there was a local conspiracy not to sell any coffee strong enough to wake up a state bureaucrat. Other than one Lebanese restaurant, you simply couldn't get espresso, and the coffee at state office buildings was watery swill that's about like what McDonald's gets while they're washing their coffee pots. The stuff at gas stations near the freeway wasn't thick enough to burn.
It's not uncommon for some kinds of startups to move to the Lake Tahoe area when they're about to make some money, so that they get Nevada's near-zero taxes instead of getting hit with California taxes. You can still drive down to civilization if you need to see people, and if you were originally Easterners instead of native Californians the idea of snow isn't scary.
Remember when there was computer hardware? Companies like Sun and Silicon Graphics and a bunch of little Motorola 680x0 workstation companies?
Yeah, that boom had ended when I moved here in the early 90s, but there was still enough interesting culture and good weather to justify moving out from the east coast, even though the Internet meant you really could work from anywhere in the world you wanted. I caught the tail end of the housing slump (which meant my house in NJ made a good down-payment on a condo out here.)
Then all the dot-com silliness happened.
Those didn't exist at my first few jobs; smokers could still smoke in the office back then :-(. But the one place I worked in the early 90s which had smokers, I'd usually join them outside for a smoke break and just stand upwind, because the break-and-socialization was an important part of the job.
Believe me, I have plenty of experience with freezing rain and cars; I currently deal with the problem by living in California. :-)
And yeah, I do prefer simple non-electronic keys - my previous two cars were Chevy vans, which meant that not only was it easy to replace keys, and carry a AAA-made plastic key in my wallet in case I got locked out, but I could also break in with a screwdriver without much trouble.
I don't need to see it in HDTV. I'd gladly go to a theater to watch the original Star Wars movie again.
I've been annoyed that I haven't been able to see the original movie since it was first in theaters back in the 70s; SW4:ANH just isn't the same thing.