My ISP uses Level 3. So whatever connection Level 3 has to Amazon(hint, direct peering). I can upload and download 100% of my line rate all the way to Germany from midwest USA, if that's any indication of the kind of "congestion" I experience.
7:32p right now
Tracing route to ec2.eu-central-1.amazonaws.com 54.239.54.28
over a maximum of 30 hops:
1 <1 ms <1 ms <1 ms pfsense.localdomain 192.168.1.1
2 1 ms 1 ms <1 ms xx.xx.xx.1
3 1 ms 1 ms 2 ms xxxx [xx.xx.xx.xx]
4 12 ms 12 ms 11 ms te-9-4.car4.Chicago1.Level3.net 4.71.102.197
13 110 ms 110 ms 110 ms ae-4-90.edge3.Frankfurt1.Level3.net 4.69.154.199
14 110 ms 110 ms 111 ms dialup-212.162.19.106.frankfurt1.mik.net 212.162.19.106
15 109 ms 110 ms 109 ms 54.239.5.82
16 110 ms 110 ms 110 ms 54.239.5.130
17 111 ms 110 ms 111 ms 54.239.6.109
18 110 ms 110 ms 110 ms 54.239.54.28
Great point. I was kind of going after a simple example that didn't involve the complexities of peering issues. In my case, my ISP does not do peering and does nearly all traffic over their trunk to Level 3, and lets Level 3 worry about routing and peering. I've only noticed one CDN, which only recently showed up, and that's to Akamai, to which I have a 1ms ping. Seems I get Windows updates from here, so it swamps my connection quite nicely. Otherwise, Netflix, YouTube, Facebook, etc, are all over their trunk to Level 3.
I could see my ISP picking up more CDNs as they increase their fiber speeds, but when I asked the person in charge of those kinds of things, he just said bandwidth is just way too cheap to worry about CDNs. And this is from an ISP that sells all Internet connections as dedicated. 30/30 for $70.
While I don't get a 99.999% guarantee for that dedicated bandwidth, they do make a best effort and keep a fairly large buffer of excess trunk bandwidth. All while charging less than the competition. Level 3 is already great at maintaining peering and keeping congestion to a minimum. This would be a lot of extra work if my ISP decided to start doing their own peering, but it could be done.
There are many things I would complain about, but an airport tracking information that you are publicly broadcasting is not one of them. You have 100% control over this. Pretty much the same thing as a server admin logging what IP addresses connect to his servers.
"voluntary" is a loaded word. If you don't work, you don't eat. It's no different than saying, take this or die. Yes, they could have worked somewhere else, but somewhere else would have been worse. How about just being "fairly" compensated? It is a complicated grey area. A company couldn't afford to pay everyone a ton of money, similar to how the record industry wants to get a slice of the money at every step of the way. But at the same time, if you add value, you should get a "fair" slice of that value added.
Hardware costs are minimal compared to all other customer support costs. The two most expensive costs of an ISP is cap-ex and customer support, both of which are drastically cheaper with fiber. And once you go fiber, it's trivial to "get the bandwidth where you want to use it". You can have your cake and eat it to. A high fiber cake:-) Keep those pipes unclogged.
A single fiber cannot handle the world's internet bandwidth
Current state of the art is 1pb/s over a single fiber, about 10x the speed of the Internet. Obviously impractical for a single fiber to connect every house in the world. The Internet is about 100tb/s right now, you can get 30tb/s over a single fiber with commercially available technology.. So 3 fibers?
And the PON systems used for homes don't even dedicate 1Gbit to each termination
WDM-PON, which is what Google Fiber uses, is 40gb/40gb with 32 lambdas of 1.25gb/1.25gb each, given each end point it's own 1.25gb/s.
You don't have a dedicated connection to a chassis with 2,000 other customers, you are PON split from a single fiber with a lot of other houses
With GPON this is the most common setup, but WDM-PON is backwards compatible with regular GPON. The most common setup is dedicated fiber back to the CO, which means an upgrade to WDM-PON is as simple as switch out the line card, then placing lambda filters on each customer's fiber, which is done back at the CO. I've called up my ISP and had them change which GPON port I was plugged into, took them about 5 minutes from the time the tech said "give me a second".
I think it's hilarious that you think that your ISP is only oversubscribing their links 2x (2,000 1Gb connections to 1Tb backhaul). That's fantasyland at the prices that residential customers pay.
I wasn't talking about the backhaul, there is no backhaul. Fiber is best described as a "Non blocking consolidator that plugs directly into the trunk". My ISP has a 3x undersubscription, in that the trunk is 3x the peak monthly peak.
Fiber can't fix bad designs, but fiber lends itself naturally to cheap, easy, and scalable designs. A single consolidator/chassis can support 2,000+ customers with 3tb+ of bandwidth. If the ISP only uses 1gb uplinks, then they're screwed anyway. But for a one time cost of $6k, you can purchase a 100gb port. They're not expensive anymore. The point is fiber makes it retardedly simple to have the entire bottle-neck be the backbone, instead of some complicated mixture of middle-mile nodes and shuffling around customers.
1gb dedicated fiber is cheaper than upgrading cable or dsl. The only way 1gb fiber is more expensive is if you do not plan to do any network upgrades. Even then, fiber will eventually be cheaper because it is about 20% cheaper to operate than cable or dsl, and operation costs are 60% of an ISP's over-all costs.
10gb over fiber for the same price as 1gb fiber, which is already cheaper than 10mb over copper, is only a matter of time. In the past 4 years, we have gone from 100gb/s 10km range over fiber to 16tb/s 600km range with no repeaters, commercially available. There have been some extremely major breakthroughs, and we can get 1pb/s over 60km or 1tb/s over 11,000km with research grade optics .
Anon already posted a link, but if the Window was limited to 64KB, I would have to be within 10ms of an end point in order to make use of my 50mb connection. This has been around for a long time. "TCP Window Scaling is implemented in Windows since Windows 2000"
When you kid is downloading some patches from Steam, Blizzard, and who knows what, all at the same time, while running BitTorrent to get the newest Linux ISOs, and remote backing up your computer.
You should not notice any issues. If you do, you don't have enough bandwidth.
Let me repeat... You should not EVER have thing think about your bandwidth or how you are using your internet connection. If you ever have to stop and think, "why is this slow", you don't have enough. You should have to micromanage what is ran and when, or who can do what at what times, etc.
We have the technology to provide every user so much bandwidth, that it's nearly impossible for them to ever run into an issue of using it all.
I stream Netflix via Level 3. Pegs my 50mb connection when doing 720p for a good 15 seconds. When talking to a higher up in my ISP, he said bandwidth from Level 3 is too cheap to care about OpenConnect. Not worth the management.
Moving around multi gigabyte files over VPN at my work? Remote backups. Bandwidth isn't expensive. You can buy 1mb of transit for $0.45 in increments of 10gb. You can get a 100gb connection for $6k/month at an IX. Couple that with Netflix OpenConnect, and free YouTube peering, and you have some pretty cheap bandwidth.
Why artificially limit our speeds? What if someone came out with a 400mhz single core CPU and tried selling it for $3k. Would you not say that's crazy in this age of tech? Same thing. Fiber is cheaper and faster, magnitudes faster.
With current technology, a single strand of fiber can handle the entire world's Internet bandwidth. Statistical multiplexing works best with large amounts of traffic, something a fiber consolidator can easily do, but copper cannot. I would rather have a 1gb fiber connection to chassis with 2,000 other customer, a 3tb/s backplane, and 1tb/s of uplink, than a 1gb coax connection with 5gb shared among 100 people, to a node that has 800 people and 20gb of uplink.
Going fiber essentially removes all choke points from the last mile, completely gets rid of the middle mile, and lets customer plug directly into the trunk. Then it's just a matter of sizing the trunk. It doesn't matter how shared it is as long as there is no congestion.
TCP_WINDOWS_SIZE can grow up to 2GB which is enough for a 10GB link with 1600ms latency. I'm not going to say that your OS will be happy about it, but that's the logical limit. As for AES 256 encryption, a modern desktop CPU can handle 100mb-300mb per core, unless you have AES-NI, then it's more like 1gb per core. OR if you're like me, you NIC supports line rate IPSEC offloading, so 4gb/s with 0% cpu overhead, assuming IPSEC and not VPN.
I have a 50/50 dedicated fiber connection with a rock solid 0.35ms ping to my ISP and a solid 8ms ping to drop box servers. Why is my connection only doing 10mb/s with DropBox and getting packet-loss, while I can use BitTorrent at 45/45 up & down at the same time and not have loss or latency? DropBox seems to have the bandwidth, but the quick bursts are wrecking havoc with my ISP's traffic shaping via their Cisco router. The way Cisco is calculating the mb/s seems to be via some sliding window, which allows a quick spike of a burst to happen in the first 1/4 of a second, but then clamps down. Because my network latency is so low, the TCP stream can ramp up really fast. Once the Cisco router clamps down on the connection, I'm already uploading nearly 100mb/s and TCP can't back off in time before loss happens.
The reason loss occurs after the clamping is because my ISP uses small buffers. They don't like buffer bloat. My max latency to my ISP before loss starts to occur is about 10ms. Since the connection is dedicated, and their trunk is sized about 3x more than peak bandwidth, it's normally not an issue.
This wouldn't be an issue if DropBox transferred the data as a single stream, but instead does a very jarring start/stop cycle, which causes my bandwidth to get very spiky. I'm thinking of enabling traffic shaping on my PFSense box, but I really don't feel like messing with it quite yet.
I guess the actual problem really is the Cisco router, but DropBox is still incredibly slow.
Let me know how well FTP scales as you add more nodes, and how it allows you to keep your data separate from other people's data while still allowing you to use their node for storage.
They are not designed to not use all of your bandwidth, it's that they can't. I've tested DropBox, and it breaks up the file into chunks and uploads them synchronously using REST calls. This meant my connection was constantly bouncing between 0% and 100%, causing bursts of packet-loss because it never gave TCP enough time to level out. BitTorrent on the other hand is great at not hosing my connection. I can run it near 100% and it will back-off as it detects latency going up, preempting the need for packet-loss to signal congestion.
Even if they weren't intentionally lying, telling investors information that is effectively false is at best being self delusional, which is professionally irresponsible.
If I didn't lock the bank vault at night, for several years in a row, and my sole job was to protect the vault, then someone broke in and stole everything, could I get away with "puffery" because I told others, with confidence, that I as protecting the vault?
How about a surgeon who is asked to do a surgery, but by no means should because he had no experience. But because the surgeon felt "optimistic", someone died. Ohh, it was just "puffery".
I can't even find any references of "Red Shift" being discredited except by fringe crazies. I have found articles talking about a certain amount of uncertainty that may need to be adjusted based on our findings of space expansion, but nothing that says it's "wrong".
When looking up Samsung products, I see few complains and many long time users. I can't say this is the case for many of the other options. Not only does Samsung compete with price for SSDs, but they have the best track record of them all, even Intel. At least the Samsung 850 has a 10 year warranty. If you're still using that SSD 10 years from now, that would be sad.
Most people I know either still live in the State or moved away for their first jobs, but have moved back to start families. Since the state covers nearly the entire cost of the Universities for in-state citizens, it doesn't really matter where they move, as long as it's still in the state.
The average GPON rollout is 3-4 years for a small ISP that has a lot of overhead. The larger you get, the more efficient you get. Even Chattanooga EPB broke even at the 3 year mark. They had completely paid off the entire fiber install and where then turning a true profit. Which is why they lowered their 1gb service from $300 to $70, because they no longer needed to pay anything off.
Slashdot Mirror
What if you don't trust root on the FTP server? Sync keeps you safe from that case, with no extra work on your end.
My ISP uses Level 3. So whatever connection Level 3 has to Amazon(hint, direct peering). I can upload and download 100% of my line rate all the way to Germany from midwest USA, if that's any indication of the kind of "congestion" I experience.
...
7:32p right now
Tracing route to ec2.eu-central-1.amazonaws.com 54.239.54.28
over a maximum of 30 hops:
1 <1 ms <1 ms <1 ms pfsense.localdomain 192.168.1.1
2 1 ms 1 ms <1 ms xx.xx.xx.1
3 1 ms 1 ms 2 ms xxxx [xx.xx.xx.xx]
4 12 ms 12 ms 11 ms te-9-4.car4.Chicago1.Level3.net 4.71.102.197
13 110 ms 110 ms 110 ms ae-4-90.edge3.Frankfurt1.Level3.net 4.69.154.199
14 110 ms 110 ms 111 ms dialup-212.162.19.106.frankfurt1.mik.net 212.162.19.106
15 109 ms 110 ms 109 ms 54.239.5.82
16 110 ms 110 ms 110 ms 54.239.5.130
17 111 ms 110 ms 111 ms 54.239.6.109
18 110 ms 110 ms 110 ms 54.239.54.28
Pinging 54.239.54.28
Packets: sent=525, rcvd=525, error=0, lost=0 (0.0% loss) in 262.110968 sec
RTTs in ms: min/avg/max/dev: 109.564 / 110.814 / 111.759 / 0.344
Great point. I was kind of going after a simple example that didn't involve the complexities of peering issues. In my case, my ISP does not do peering and does nearly all traffic over their trunk to Level 3, and lets Level 3 worry about routing and peering. I've only noticed one CDN, which only recently showed up, and that's to Akamai, to which I have a 1ms ping. Seems I get Windows updates from here, so it swamps my connection quite nicely. Otherwise, Netflix, YouTube, Facebook, etc, are all over their trunk to Level 3.
I could see my ISP picking up more CDNs as they increase their fiber speeds, but when I asked the person in charge of those kinds of things, he just said bandwidth is just way too cheap to worry about CDNs. And this is from an ISP that sells all Internet connections as dedicated. 30/30 for $70.
While I don't get a 99.999% guarantee for that dedicated bandwidth, they do make a best effort and keep a fairly large buffer of excess trunk bandwidth. All while charging less than the competition. Level 3 is already great at maintaining peering and keeping congestion to a minimum. This would be a lot of extra work if my ISP decided to start doing their own peering, but it could be done.
There are many things I would complain about, but an airport tracking information that you are publicly broadcasting is not one of them. You have 100% control over this. Pretty much the same thing as a server admin logging what IP addresses connect to his servers.
"voluntary" is a loaded word. If you don't work, you don't eat. It's no different than saying, take this or die. Yes, they could have worked somewhere else, but somewhere else would have been worse. How about just being "fairly" compensated? It is a complicated grey area. A company couldn't afford to pay everyone a ton of money, similar to how the record industry wants to get a slice of the money at every step of the way. But at the same time, if you add value, you should get a "fair" slice of that value added.
Meant "op-ex" not "cap-ex"
Hardware costs are minimal compared to all other customer support costs. The two most expensive costs of an ISP is cap-ex and customer support, both of which are drastically cheaper with fiber. And once you go fiber, it's trivial to "get the bandwidth where you want to use it". You can have your cake and eat it to. A high fiber cake :-) Keep those pipes unclogged.
Who said anything about my home system? Only work computers can VPN in, but I can take my laptop home and work remotely.
A single fiber cannot handle the world's internet bandwidth
Current state of the art is 1pb/s over a single fiber, about 10x the speed of the Internet. Obviously impractical for a single fiber to connect every house in the world. The Internet is about 100tb/s right now, you can get 30tb/s over a single fiber with commercially available technology.. So 3 fibers?
And the PON systems used for homes don't even dedicate 1Gbit to each termination
WDM-PON, which is what Google Fiber uses, is 40gb/40gb with 32 lambdas of 1.25gb/1.25gb each, given each end point it's own 1.25gb/s.
You don't have a dedicated connection to a chassis with 2,000 other customers, you are PON split from a single fiber with a lot of other houses
With GPON this is the most common setup, but WDM-PON is backwards compatible with regular GPON. The most common setup is dedicated fiber back to the CO, which means an upgrade to WDM-PON is as simple as switch out the line card, then placing lambda filters on each customer's fiber, which is done back at the CO. I've called up my ISP and had them change which GPON port I was plugged into, took them about 5 minutes from the time the tech said "give me a second".
I think it's hilarious that you think that your ISP is only oversubscribing their links 2x (2,000 1Gb connections to 1Tb backhaul). That's fantasyland at the prices that residential customers pay.
I wasn't talking about the backhaul, there is no backhaul. Fiber is best described as a "Non blocking consolidator that plugs directly into the trunk". My ISP has a 3x undersubscription, in that the trunk is 3x the peak monthly peak.
Fiber can't fix bad designs, but fiber lends itself naturally to cheap, easy, and scalable designs. A single consolidator/chassis can support 2,000+ customers with 3tb+ of bandwidth. If the ISP only uses 1gb uplinks, then they're screwed anyway. But for a one time cost of $6k, you can purchase a 100gb port. They're not expensive anymore. The point is fiber makes it retardedly simple to have the entire bottle-neck be the backbone, instead of some complicated mixture of middle-mile nodes and shuffling around customers.
but why would anyone assume it should be free
1gb dedicated fiber is cheaper than upgrading cable or dsl. The only way 1gb fiber is more expensive is if you do not plan to do any network upgrades. Even then, fiber will eventually be cheaper because it is about 20% cheaper to operate than cable or dsl, and operation costs are 60% of an ISP's over-all costs.
10gb over fiber for the same price as 1gb fiber, which is already cheaper than 10mb over copper, is only a matter of time. In the past 4 years, we have gone from 100gb/s 10km range over fiber to 16tb/s 600km range with no repeaters, commercially available. There have been some extremely major breakthroughs, and we can get 1pb/s over 60km or 1tb/s over 11,000km with research grade optics .
Anon already posted a link, but if the Window was limited to 64KB, I would have to be within 10ms of an end point in order to make use of my 50mb connection. This has been around for a long time. "TCP Window Scaling is implemented in Windows since Windows 2000"
When you kid is downloading some patches from Steam, Blizzard, and who knows what, all at the same time, while running BitTorrent to get the newest Linux ISOs, and remote backing up your computer.
You should not notice any issues. If you do, you don't have enough bandwidth.
Let me repeat... You should not EVER have thing think about your bandwidth or how you are using your internet connection. If you ever have to stop and think, "why is this slow", you don't have enough. You should have to micromanage what is ran and when, or who can do what at what times, etc.
We have the technology to provide every user so much bandwidth, that it's nearly impossible for them to ever run into an issue of using it all.
I stream Netflix via Level 3. Pegs my 50mb connection when doing 720p for a good 15 seconds. When talking to a higher up in my ISP, he said bandwidth from Level 3 is too cheap to care about OpenConnect. Not worth the management.
Moving around multi gigabyte files over VPN at my work? Remote backups. Bandwidth isn't expensive. You can buy 1mb of transit for $0.45 in increments of 10gb. You can get a 100gb connection for $6k/month at an IX. Couple that with Netflix OpenConnect, and free YouTube peering, and you have some pretty cheap bandwidth.
Why artificially limit our speeds? What if someone came out with a 400mhz single core CPU and tried selling it for $3k. Would you not say that's crazy in this age of tech? Same thing. Fiber is cheaper and faster, magnitudes faster.
With current technology, a single strand of fiber can handle the entire world's Internet bandwidth. Statistical multiplexing works best with large amounts of traffic, something a fiber consolidator can easily do, but copper cannot. I would rather have a 1gb fiber connection to chassis with 2,000 other customer, a 3tb/s backplane, and 1tb/s of uplink, than a 1gb coax connection with 5gb shared among 100 people, to a node that has 800 people and 20gb of uplink.
Going fiber essentially removes all choke points from the last mile, completely gets rid of the middle mile, and lets customer plug directly into the trunk. Then it's just a matter of sizing the trunk. It doesn't matter how shared it is as long as there is no congestion.
TCP_WINDOWS_SIZE can grow up to 2GB which is enough for a 10GB link with 1600ms latency. I'm not going to say that your OS will be happy about it, but that's the logical limit. As for AES 256 encryption, a modern desktop CPU can handle 100mb-300mb per core, unless you have AES-NI, then it's more like 1gb per core. OR if you're like me, you NIC supports line rate IPSEC offloading, so 4gb/s with 0% cpu overhead, assuming IPSEC and not VPN.
I have a 50/50 dedicated fiber connection with a rock solid 0.35ms ping to my ISP and a solid 8ms ping to drop box servers. Why is my connection only doing 10mb/s with DropBox and getting packet-loss, while I can use BitTorrent at 45/45 up & down at the same time and not have loss or latency? DropBox seems to have the bandwidth, but the quick bursts are wrecking havoc with my ISP's traffic shaping via their Cisco router. The way Cisco is calculating the mb/s seems to be via some sliding window, which allows a quick spike of a burst to happen in the first 1/4 of a second, but then clamps down. Because my network latency is so low, the TCP stream can ramp up really fast. Once the Cisco router clamps down on the connection, I'm already uploading nearly 100mb/s and TCP can't back off in time before loss happens.
The reason loss occurs after the clamping is because my ISP uses small buffers. They don't like buffer bloat. My max latency to my ISP before loss starts to occur is about 10ms. Since the connection is dedicated, and their trunk is sized about 3x more than peak bandwidth, it's normally not an issue.
This wouldn't be an issue if DropBox transferred the data as a single stream, but instead does a very jarring start/stop cycle, which causes my bandwidth to get very spiky. I'm thinking of enabling traffic shaping on my PFSense box, but I really don't feel like messing with it quite yet.
I guess the actual problem really is the Cisco router, but DropBox is still incredibly slow.
Let me know how well FTP scales as you add more nodes, and how it allows you to keep your data separate from other people's data while still allowing you to use their node for storage.
They are not designed to not use all of your bandwidth, it's that they can't. I've tested DropBox, and it breaks up the file into chunks and uploads them synchronously using REST calls. This meant my connection was constantly bouncing between 0% and 100%, causing bursts of packet-loss because it never gave TCP enough time to level out. BitTorrent on the other hand is great at not hosing my connection. I can run it near 100% and it will back-off as it detects latency going up, preempting the need for packet-loss to signal congestion.
Even if they weren't intentionally lying, telling investors information that is effectively false is at best being self delusional, which is professionally irresponsible.
If I didn't lock the bank vault at night, for several years in a row, and my sole job was to protect the vault, then someone broke in and stole everything, could I get away with "puffery" because I told others, with confidence, that I as protecting the vault?
How about a surgeon who is asked to do a surgery, but by no means should because he had no experience. But because the surgeon felt "optimistic", someone died. Ohh, it was just "puffery".
Call it what it is, willfully ignorant lying.
I can't even find any references of "Red Shift" being discredited except by fringe crazies. I have found articles talking about a certain amount of uncertainty that may need to be adjusted based on our findings of space expansion, but nothing that says it's "wrong".
When looking up Samsung products, I see few complains and many long time users. I can't say this is the case for many of the other options. Not only does Samsung compete with price for SSDs, but they have the best track record of them all, even Intel. At least the Samsung 850 has a 10 year warranty. If you're still using that SSD 10 years from now, that would be sad.
"and shot a soldier with a rifle."
Is it ambiguous for others of whether the soldier was shot with a rifle or the soldier with a rifle was shot?
When I read it in my head, I naturally want to place a pause after "soldier", so I assume it should be more like "and shot a soldier, with a rifle".
Most people I know either still live in the State or moved away for their first jobs, but have moved back to start families. Since the state covers nearly the entire cost of the Universities for in-state citizens, it doesn't really matter where they move, as long as it's still in the state.
The average GPON rollout is 3-4 years for a small ISP that has a lot of overhead. The larger you get, the more efficient you get. Even Chattanooga EPB broke even at the 3 year mark. They had completely paid off the entire fiber install and where then turning a true profit. Which is why they lowered their 1gb service from $300 to $70, because they no longer needed to pay anything off.