Are you sure? 400m in fiber is about 2.67 usec for light.
299,792,458m/s/1,000,000usec=299.792458m
400/299.792458m=1.3342563807926081983023068578997
Light in fiber is closer to.55c, which is close enough to.5
1.33*2=2.67us to travel 400m
Queuing time would be in addition to the shortening.
The positive thing is that they'll lay all of this redundant ultra-low latency fiber, go bust at some point when it's finally regulated, then everyone will make use of lower pings.
"USB 3.0 is backwards compatible" Not for long. USB is near its limit, it will soon have to choose between backwards compatibility or speed. Once that point is reached, there will be no benefit to stick with USB. Anyway, USB is cheap enough to keep around along side of TB.
Benchmarks have shown TB getting an effective 10Gb both directions at the same time with several devices at the same time, at the data layer, so a true "effective" throughput.
unless Mary looks at A and B and decides she doesn't like the GPL, re-invents the wheel and still creates C or C doesn't get created at all.
Or another case for BSD. Mary gets a contract from a company which says their few lines of proprietary code must remain secret. So Mary creates C and C.1(proprietary code). She releases C upstream for everyone else to use and keeps C.1 secret to honor the contract. All of the upstream people still benefit from the bulk of the work.
Remember, the biggest incentive to release your code upstream is that you no longer have to maintain it, which is much much much cheaper as your code would otherwise diverge from the trunk.
Claiming losses on sales that never happened in the first place is like this.
Say littering gets you a $100 fine. Say if you spent all day littering in front of a police station, you could rack up $5000 in fines. This means if you don't litter, it's like saving $5000/day, which means people who don't litter at all are "saving" more than $1.8mil/year. They're all rich!
One might say "taking credit" is the only real issue copyright should protect. copy someone else's work all you want, but don't you take credit for creating it. One could say BSD is the most free before detracting from the creator.
Bandwidth is cheaper than QoS, except at the edge. Once you hit the ISP, all bets are off. Huge bandwidth aggregation makes QoS a nightmare. It's cheaper just to use dumb hardware and team more 10Gb/40Gb ports together. Even at the core trunk, bandwidth is still cheaper than QoS.
Charging for QoS is just another way to "Tax" people above and beyond.
What they are giving you is a discount that is reflected by the technical reality that they can transmit video to you over their own network for a lower cost than access to services on the internet at large.
You're the one not in touch with reality. Level3 is the primary CDN for Netflix, This means all peers/customers with Level3 get Netflix bandwidth for FREE. Comcast is not only a peer with Level3, but they use Level3 and one of their many links to the internet. Netflix doesn't cost Comcast any more than their own internal services.
There is a distinct difference between charging "less" and artificially adding a price to a competitor's service.
eg. Netflix $8/month. If Comcast could offer a competitive service for $6/month because of the money the same from bandwidth, that's fine
What Comcast is doing is doing is charging customers *extra* on top of Netflix's service through caps and giving priority to Comcast's own services.
If I owned road tolls in Illinois and a car dealership, then only stopped cars that weren't from my car lot and charged them money to use my roads. But not any roads, but the *only* roads in the city because I had a monopoly.
You just mixed a lot of online and offline ideas together. You can't throttle offline cracking, unless you ask the cracker "pretty please with sugar on top".
I haven't heard much of online cracking being an issue because of what you said; one simply can't effectively attempt many passwords without causing a DoS.
"The work itself is very nice, MD5 hashes can be cracked quickly in massive parallel on GPU hardware. This only matters after the hashes have already been stolen." This part is true.
Locking down accounts sounds great until you think of what would happen. Take gmail. Some large bot owner decides to make 20 attempts on every account. Soon no one in the world can access their gmail because their accounts are constantly locked. How long until Google decides to remove account locking on failed attempts?
Some people have mentioned 2 factor auth. One could also filter by location when deciding to block attempts. They could also setup a special URL that is unique per person and can be re-generated. Instead of the normal login page, you get this on-the-side page that is immune to account locks.
I'm sure there are lots of ideas, each with a trade off.
One does have to add that the password would have to be focused on, which doesn't happen with database dumps like these, and it would cost a pretty penny. 10 char password with mixed/etc would take on average 1 year when doing 1 tril checks/sec. To break within an hour, one would need access to ~80-800 peta-flops of compute.
Definitely something someone targeting a password could do, but it isn't cheap.
I would -- I would be bored to death if I had nothing to do, and my job is the same as my hobby.
Yep. Makes life much more enjoyable. What do I do in my "fun" time? I get better at my job. What do I think when I wake up? "I can't wait to test out my project in production!"
Actually, it's a bit different than what you're saying.
Netflix used to use Akamai, but then Level3 under-bid them. Akamai is a local cache CDN while Level3 is a Tier1 back-bone that uses it's massive cheap trunks to compete. It is cheaper for Level3 to deliver Netflix over the back-bone than for Akamai to cache the data locally. Akamai is still useful because not all Netflix customer's ISPs have peering with L3.
Because Level3 is acting as a CDN in the case of Netflix, this means all of the cost of Netflix traffic via L3 is borne by Netflix. If Netflix traffic is hitting an L3 peer, that peer will not have Netflix traffic count towards them.
So in the case of my ISP who uses Level3 as their upstream, Netflix traffic is free and does not count towards their 95th percentile billing. This is good because on average, ~30% of an ISPs bandwidth during peak is Netflix.
As long as services like netflix entice people to all peg their connections at the same time of day
I didn't realize 4Mb streams are "pegging" 30Mb connections.
According to Level3 engineer blogs, the cheapest and most effective form of QoS is bandwidth. At least for big players like L3, upgrading connections is much cheaper than having hardware that can route and QoS at the speeds they handle. L3 also claims the prices they offer their clients are low enough that bandwidth is almost always cheaper than QoS.
Core backbone has no problem. Prices are cheap and lots of competition, the issue is the last mile. Local caching is just an efficiency thing. With the HUGE growth of streaming, local caches will take some of the bite out of the growth pain. There are quite a few long-haul high speed 500Gb+ fiber techs around the corner, but they're not quite here yet. Your only options for high speed trunks is expensive equipment or lots of 10Gb fiber. CDNs are going to give us that breathing room while waiting for the new tech to become affordable.
Of course Netflix pays for it's bandwidth, he wasn't arguing that. Netflix pays for its connection to the internet and the customer pays for their connection. As long as everyone pays for their connection, there is no reason to say Netflix is "hogging" the bandwidth, which is what Comcast is trying to say.
Slashdot Mirror
Are you sure? 400m in fiber is about 2.67 usec for light.
.55c, which is close enough to .5
299,792,458m/s/1,000,000usec=299.792458m
400/299.792458m=1.3342563807926081983023068578997
Light in fiber is closer to
1.33*2=2.67us to travel 400m
Queuing time would be in addition to the shortening.
The positive thing is that they'll lay all of this redundant ultra-low latency fiber, go bust at some point when it's finally regulated, then everyone will make use of lower pings.
"USB 3.0 is backwards compatible" Not for long. USB is near its limit, it will soon have to choose between backwards compatibility or speed. Once that point is reached, there will be no benefit to stick with USB. Anyway, USB is cheap enough to keep around along side of TB.
Benchmarks have shown TB getting an effective 10Gb both directions at the same time with several devices at the same time, at the data layer, so a true "effective" throughput.
I hope 10 comes out next year some time. I really want to see PFSense using Netmap in the next 3 years.
unless Mary looks at A and B and decides she doesn't like the GPL, re-invents the wheel and still creates C or C doesn't get created at all.
Or another case for BSD. Mary gets a contract from a company which says their few lines of proprietary code must remain secret. So Mary creates C and C.1(proprietary code). She releases C upstream for everyone else to use and keeps C.1 secret to honor the contract. All of the upstream people still benefit from the bulk of the work.
Remember, the biggest incentive to release your code upstream is that you no longer have to maintain it, which is much much much cheaper as your code would otherwise diverge from the trunk.
I came across this analogy.. paraphrased.
Claiming losses on sales that never happened in the first place is like this.
Say littering gets you a $100 fine. Say if you spent all day littering in front of a police station, you could rack up $5000 in fines. This means if you don't litter, it's like saving $5000/day, which means people who don't litter at all are "saving" more than $1.8mil/year. They're all rich!
Apple has given back a lot to BSD because GPL didn't want any of it unless it could get all of it.
Perfect is the enemy of good. GPL is a bit idealistic. Great ideal, can't argue that.
Better check your network stack on Linux. I'm sure it has a BSD logo on it.
One might say "taking credit" is the only real issue copyright should protect. copy someone else's work all you want, but don't you take credit for creating it. One could say BSD is the most free before detracting from the creator.
They aren't targeting people, they're targeting a certain feature.
Bandwidth is cheaper than QoS, except at the edge. Once you hit the ISP, all bets are off. Huge bandwidth aggregation makes QoS a nightmare. It's cheaper just to use dumb hardware and team more 10Gb/40Gb ports together. Even at the core trunk, bandwidth is still cheaper than QoS.
Charging for QoS is just another way to "Tax" people above and beyond.
What they are giving you is a discount that is reflected by the technical reality that they can transmit video to you over their own network for a lower cost than access to services on the internet at large.
You're the one not in touch with reality. Level3 is the primary CDN for Netflix, This means all peers/customers with Level3 get Netflix bandwidth for FREE. Comcast is not only a peer with Level3, but they use Level3 and one of their many links to the internet. Netflix doesn't cost Comcast any more than their own internal services.
There is a distinct difference between charging "less" and artificially adding a price to a competitor's service.
eg. Netflix $8/month. If Comcast could offer a competitive service for $6/month because of the money the same from bandwidth, that's fine
What Comcast is doing is doing is charging customers *extra* on top of Netflix's service through caps and giving priority to Comcast's own services.
If I owned road tolls in Illinois and a car dealership, then only stopped cars that weren't from my car lot and charged them money to use my roads. But not any roads, but the *only* roads in the city because I had a monopoly.
I could care less where you come from, I welcome all good-willed hard workers with open arms. Welcome to the USA.
End to end quality
Ever get an "all circuits are busy" message? Just different granularity during congestion. Instead of per packet, it's per connection.
80/20 rule for life!
You just mixed a lot of online and offline ideas together. You can't throttle offline cracking, unless you ask the cracker "pretty please with sugar on top".
I haven't heard much of online cracking being an issue because of what you said; one simply can't effectively attempt many passwords without causing a DoS.
"The work itself is very nice, MD5 hashes can be cracked quickly in massive parallel on GPU hardware. This only matters after the hashes have already been stolen." This part is true.
Locking down accounts sounds great until you think of what would happen. Take gmail. Some large bot owner decides to make 20 attempts on every account. Soon no one in the world can access their gmail because their accounts are constantly locked. How long until Google decides to remove account locking on failed attempts?
Some people have mentioned 2 factor auth. One could also filter by location when deciding to block attempts. They could also setup a special URL that is unique per person and can be re-generated. Instead of the normal login page, you get this on-the-side page that is immune to account locks.
I'm sure there are lots of ideas, each with a trade off.
One does have to add that the password would have to be focused on, which doesn't happen with database dumps like these, and it would cost a pretty penny. 10 char password with mixed/etc would take on average 1 year when doing 1 tril checks/sec. To break within an hour, one would need access to ~80-800 peta-flops of compute. Definitely something someone targeting a password could do, but it isn't cheap.
But instead of 4 hours to crack 2mil hashes, he would have spent 4 hours per hash for 2mil hashes. It makes a small difference.
I would -- I would be bored to death if I had nothing to do, and my job is the same as my hobby.
Yep. Makes life much more enjoyable. What do I do in my "fun" time? I get better at my job. What do I think when I wake up? "I can't wait to test out my project in production!"
Netflix used to use Akamai, but then Level3 under-bid them. Akamai is a local cache CDN while Level3 is a Tier1 back-bone that uses it's massive cheap trunks to compete. It is cheaper for Level3 to deliver Netflix over the back-bone than for Akamai to cache the data locally. Akamai is still useful because not all Netflix customer's ISPs have peering with L3.
Because Level3 is acting as a CDN in the case of Netflix, this means all of the cost of Netflix traffic via L3 is borne by Netflix. If Netflix traffic is hitting an L3 peer, that peer will not have Netflix traffic count towards them.
So in the case of my ISP who uses Level3 as their upstream, Netflix traffic is free and does not count towards their 95th percentile billing. This is good because on average, ~30% of an ISPs bandwidth during peak is Netflix.
As long as services like netflix entice people to all peg their connections at the same time of day
I didn't realize 4Mb streams are "pegging" 30Mb connections.
According to Level3 engineer blogs, the cheapest and most effective form of QoS is bandwidth. At least for big players like L3, upgrading connections is much cheaper than having hardware that can route and QoS at the speeds they handle. L3 also claims the prices they offer their clients are low enough that bandwidth is almost always cheaper than QoS.
There has always been many many more consumers than producers of content. Cache that read-only static data closer to the consumer.
Core backbone has no problem. Prices are cheap and lots of competition, the issue is the last mile. Local caching is just an efficiency thing. With the HUGE growth of streaming, local caches will take some of the bite out of the growth pain. There are quite a few long-haul high speed 500Gb+ fiber techs around the corner, but they're not quite here yet. Your only options for high speed trunks is expensive equipment or lots of 10Gb fiber. CDNs are going to give us that breathing room while waiting for the new tech to become affordable.
Of course Netflix pays for it's bandwidth, he wasn't arguing that. Netflix pays for its connection to the internet and the customer pays for their connection. As long as everyone pays for their connection, there is no reason to say Netflix is "hogging" the bandwidth, which is what Comcast is trying to say.