I pay less for guaranteed speeds. 20/20 for $20/m, 70/70 for $34, and 100/100 for $45. All guaranteed rates 1 hop Level 3(tier 1 ISP). Fast high quality internet is cheaper than the crap you pay for.
The role of analysts and architects are very important, but any great programmer should be able to handle those roles, you don't need someone 100% dedicated to these positions. The bigger issue is that most programmers are not great programmers and can't handle those roles. Few people can take a holistic view of the entire system and make the parts fit elegantly, and instead create a spaghetti mess.
From what I've read, these companies are hyper-competitive and only accept the best. If anything, they're trying to flood the market to free up the best from smaller companies so they can hoard them.
Well, it is a privately own ISP that started back when telegraph was all the rage. They also openly turn down government grants and loans. We're also in a small town with a high unemployment rate and low median income. Even in this horrible situation, they still manage to sell uncapped dedicated symmetrical internet for a fraction of the price of Charter or AT&T. My $45 home connection has magnitudes less jitter and a fraction of latency to the rest of the internet than my job's enterprise 10Gb connection to Charter. They only reason they stick with Charter is they are cheaper for enterprise bandwidth.
Forgive me if I assume they know what they're doing.
Knowledge does help, but some people can walk into a problem with virtually no related knowledge and still come up with good answers quickly. In these situations, it seems to be someone's ability to simplify a problem, use reason, and ask good questions to quickly find their missing knowledge.
Google for Cake and fq_CoDel(cake's ancestor). Stateless AQMs that maximize bandwidth and minimize latency. Virtually no configuration and does 99% of what everyone wants.
An ISP must be incredibly wasteful with money if customer service doesn't cost more. The average real world costs of fiber is nearly 1/5th that of copper. Maybe your ISPs were using copper? Even nodes? Nodes in the field are expensive. The cheapest design is a flat model where customers plug into fiber aggregators that plug directly into the local core router.
I've read article after article over the years where they said incumbent ISPs are incredibly wasteful and bandwidth and infrastructure is relatively cheap over the lifetime of a customer. When talking to an ISP network admin, he told me they tried doing QoS and traffic shaping, but issues always cropped up and customer would complain because of poor performance. The cost of handling customers was too expensive, so they just went 100% dedicated. Every customer is guaranteed to have a congestion free, with no caps, traffic shaping, or QoS. Just pure unfettered bandwidth to brute force the issue. Costs went down, customer satisfaction went up.
The only expensive part of correctly designing a network is the up-front cost. But after 3-8 years, that cost is dwarfed by ongoing costs.
Dark matter has also been discovered in galactic voids. This would never be explainable by the electric force, which requires normal matter to be present.
They can't afford a $30 Raspberry Pi, but they can afford a $600 iPhone? I know plenty of families making less than $30k/year household income, have kids, a big screen TV, deluxe cable, smart phones, a few game consoles, but no computers. Don't tell me it's because they can't "afford" a computer.
All obvious candidates have been accounted for and failed. It comes down to this. Whatever Dark Matter is, it can't be baryonic. It has to be slow moving, which means massive, it can't interact with itself easily, and it can't interact with the electromagnetic force. Of all known matter, none of them foot the bill without some serious modifications that are more ludicrous than Dark Matter itself.
Depends on your definition of "directly". Have you touched the Sun? Ohh, you're just using instruments and measuring results, just like Dark Matter. We only have indirect evidence that the Sun exists.
Dark Matter is the simplest because if it is not real, then we have to ditch all of current physics. It means Relativity is wrong, and not just a little wrong, but incredibly wrong. And that's just the tip of the iceburg. It also implies that all of our measurements are wrong in the first place. Can you see the logical implication of assuming our measurements of the rotations are correct, but everything else is wrong? It really is a can of worms. Of course nothing is 100%, it's just the current best.
The alternative to Dark Matter is crazier than Dark Matter. The lesser of two evils if you will. There is a lot of mounting evidence. After decades of work, someone finally got MOND to work, but it required a matter that has not been detected and is not the same as any other known matter. They reinvented Dark Matter.
There is still some research going on that some of the leading names in quantum physics are hopeful with. They take two entangled photons, A and B, send A at another sensor a long distance away, then entangle B with C, where C is known. Because when you measure an entangled photon, the other photon becomes the opposite. So when C is entangled with B, then B because anti-C, but because B is also entangled with A, A because anti-B, which is the same as C. But in this process, C is randomized and loses its original information.
I guess preliminary tests show this to probably work, but they need more samples and it is very hard to do.
You must be from a planet where Raspberry PIs cost $1.5k because there's no competition. But then you still claim we're getting a good deal because it's still faster and cheaper than an 8086.
Instead of providing the bandwidth paid for, people who roll-over for abusive ISPs by using less bandwidth, should get rewarded by getting priority over those who make use of their paid service. My argument is not against small ISPs, but normal sized ones where bandwidth is a rounding error compared to other costs.
AQMs solve the issues people have with QoS and traffic shaping. Instead of doing strict prioritizing, sprase flows get strict priority and heavy flows get bandwidth evenly distributed. http://www.bufferbloat.net/pro...
Customer service costs more than bandwidth or infrastructure combined. You may want to revisit your logic about wasting huge amounts of money trying to screw with customers.
Many people are too blind to nuances to make proper use of their experience. I've seen it where we needed a new SAN device for a data warehouse. We told them specifically that sequential performance is more important than random access. What did they get? Some $200k high IOPs device that has great multi-LUN performance and poor few large LUN sequential access performance.
People miss-use best practices all the time because they don't understand the "Why" aspect.
I pay $45/m for a dedicated 100/100 fiber connection to my home here in Midwest USA. My max sustained 1min average bandwidth is 99.9Mb/s. My typical sustained bandwidth when downloading or uploading torrents is 99.5Mb/s, and my min is about 99.25Mb/s. These aren't iPerf tests, but HTTPS/FTP/Torrent. Without doing any traffic shaping, while saturating my connection, my ping never jitters more than 30ms. When I do my own traffic shaping and rate limit to 99.9Mb/s, my ping never jitters more than 1ms.
My max ping to any datacenter in the world is about 250ms and under 5ms of jitter. Upstream provider of my ISP is Level 3 Comm. My ISP guarantees 1:1 internal bandwidth and typically has about 6:1 of trunk bandwidth based on 95th percentile. They are under-subscribed.
I actually did a 120Mb/s external DOS stress test on my connection. The ping to my ISP stayed under 30ms the entire time, but I did have about 20% packetloss. They use a fair-queuing AQM for both up and down to fight bufferbloat.
I've done 1 week long pings to datacenters in LA, London, and Germany. All of them over a 5,000 mile round trip. Less than 5ms of jitter to Europe and under 1ms to LA, and 0.0001% packetloss. My friend was an admin at a datacenter where they had dual DS3 connections to AT&T. He got a 200ms ping to Hawaii and some nasty jitter. I told him I got a 140ms ping and virtually no jitter. He didn't believe me until I got some samples. Even Japan is only 160ms.
The first two generations of stars we HUGE and died quickly, destroying any planets near them. That's the whole issue. Our planet is part of the first generation of planets that would last long enough for intelligent life to evolve.
Slashdot Mirror
I pay less for guaranteed speeds. 20/20 for $20/m, 70/70 for $34, and 100/100 for $45. All guaranteed rates 1 hop Level 3(tier 1 ISP). Fast high quality internet is cheaper than the crap you pay for.
The role of analysts and architects are very important, but any great programmer should be able to handle those roles, you don't need someone 100% dedicated to these positions. The bigger issue is that most programmers are not great programmers and can't handle those roles. Few people can take a holistic view of the entire system and make the parts fit elegantly, and instead create a spaghetti mess.
From what I've read, these companies are hyper-competitive and only accept the best. If anything, they're trying to flood the market to free up the best from smaller companies so they can hoard them.
Well, it is a privately own ISP that started back when telegraph was all the rage. They also openly turn down government grants and loans. We're also in a small town with a high unemployment rate and low median income. Even in this horrible situation, they still manage to sell uncapped dedicated symmetrical internet for a fraction of the price of Charter or AT&T. My $45 home connection has magnitudes less jitter and a fraction of latency to the rest of the internet than my job's enterprise 10Gb connection to Charter. They only reason they stick with Charter is they are cheaper for enterprise bandwidth.
Forgive me if I assume they know what they're doing.
I think the government should be guilty until proven innocent.
Knowledge does help, but some people can walk into a problem with virtually no related knowledge and still come up with good answers quickly. In these situations, it seems to be someone's ability to simplify a problem, use reason, and ask good questions to quickly find their missing knowledge.
On average, P2P is now a small percentage of download bandwidth, but you may have a bias with lots of college students.
Google for Cake and fq_CoDel(cake's ancestor). Stateless AQMs that maximize bandwidth and minimize latency. Virtually no configuration and does 99% of what everyone wants.
An ISP must be incredibly wasteful with money if customer service doesn't cost more. The average real world costs of fiber is nearly 1/5th that of copper. Maybe your ISPs were using copper? Even nodes? Nodes in the field are expensive. The cheapest design is a flat model where customers plug into fiber aggregators that plug directly into the local core router.
I've read article after article over the years where they said incumbent ISPs are incredibly wasteful and bandwidth and infrastructure is relatively cheap over the lifetime of a customer. When talking to an ISP network admin, he told me they tried doing QoS and traffic shaping, but issues always cropped up and customer would complain because of poor performance. The cost of handling customers was too expensive, so they just went 100% dedicated. Every customer is guaranteed to have a congestion free, with no caps, traffic shaping, or QoS. Just pure unfettered bandwidth to brute force the issue. Costs went down, customer satisfaction went up.
The only expensive part of correctly designing a network is the up-front cost. But after 3-8 years, that cost is dwarfed by ongoing costs.
Dark matter has also been discovered in galactic voids. This would never be explainable by the electric force, which requires normal matter to be present.
They can't afford a $30 Raspberry Pi, but they can afford a $600 iPhone? I know plenty of families making less than $30k/year household income, have kids, a big screen TV, deluxe cable, smart phones, a few game consoles, but no computers. Don't tell me it's because they can't "afford" a computer.
All obvious candidates have been accounted for and failed. It comes down to this. Whatever Dark Matter is, it can't be baryonic. It has to be slow moving, which means massive, it can't interact with itself easily, and it can't interact with the electromagnetic force. Of all known matter, none of them foot the bill without some serious modifications that are more ludicrous than Dark Matter itself.
Depends on your definition of "directly". Have you touched the Sun? Ohh, you're just using instruments and measuring results, just like Dark Matter. We only have indirect evidence that the Sun exists.
There is a difference between a refractive index and a warping of space time.
Dark Matter is the simplest because if it is not real, then we have to ditch all of current physics. It means Relativity is wrong, and not just a little wrong, but incredibly wrong. And that's just the tip of the iceburg. It also implies that all of our measurements are wrong in the first place. Can you see the logical implication of assuming our measurements of the rotations are correct, but everything else is wrong? It really is a can of worms. Of course nothing is 100%, it's just the current best.
The alternative to Dark Matter is crazier than Dark Matter. The lesser of two evils if you will. There is a lot of mounting evidence. After decades of work, someone finally got MOND to work, but it required a matter that has not been detected and is not the same as any other known matter. They reinvented Dark Matter.
There is still some research going on that some of the leading names in quantum physics are hopeful with. They take two entangled photons, A and B, send A at another sensor a long distance away, then entangle B with C, where C is known. Because when you measure an entangled photon, the other photon becomes the opposite. So when C is entangled with B, then B because anti-C, but because B is also entangled with A, A because anti-B, which is the same as C. But in this process, C is randomized and loses its original information.
I guess preliminary tests show this to probably work, but they need more samples and it is very hard to do.
You must be from a planet where Raspberry PIs cost $1.5k because there's no competition. But then you still claim we're getting a good deal because it's still faster and cheaper than an 8086.
Instead of providing the bandwidth paid for, people who roll-over for abusive ISPs by using less bandwidth, should get rewarded by getting priority over those who make use of their paid service. My argument is not against small ISPs, but normal sized ones where bandwidth is a rounding error compared to other costs.
AQMs solve the issues people have with QoS and traffic shaping. Instead of doing strict prioritizing, sprase flows get strict priority and heavy flows get bandwidth evenly distributed. http://www.bufferbloat.net/pro...
Customer service costs more than bandwidth or infrastructure combined. You may want to revisit your logic about wasting huge amounts of money trying to screw with customers.
You stated it as a fact. All I have to do is fine an exception to make your claim false.
Many people are too blind to nuances to make proper use of their experience. I've seen it where we needed a new SAN device for a data warehouse. We told them specifically that sequential performance is more important than random access. What did they get? Some $200k high IOPs device that has great multi-LUN performance and poor few large LUN sequential access performance.
People miss-use best practices all the time because they don't understand the "Why" aspect.
I pay $45/m for a dedicated 100/100 fiber connection to my home here in Midwest USA. My max sustained 1min average bandwidth is 99.9Mb/s. My typical sustained bandwidth when downloading or uploading torrents is 99.5Mb/s, and my min is about 99.25Mb/s. These aren't iPerf tests, but HTTPS/FTP/Torrent. Without doing any traffic shaping, while saturating my connection, my ping never jitters more than 30ms. When I do my own traffic shaping and rate limit to 99.9Mb/s, my ping never jitters more than 1ms.
My max ping to any datacenter in the world is about 250ms and under 5ms of jitter. Upstream provider of my ISP is Level 3 Comm. My ISP guarantees 1:1 internal bandwidth and typically has about 6:1 of trunk bandwidth based on 95th percentile. They are under-subscribed.
I actually did a 120Mb/s external DOS stress test on my connection. The ping to my ISP stayed under 30ms the entire time, but I did have about 20% packetloss. They use a fair-queuing AQM for both up and down to fight bufferbloat.
I've done 1 week long pings to datacenters in LA, London, and Germany. All of them over a 5,000 mile round trip. Less than 5ms of jitter to Europe and under 1ms to LA, and 0.0001% packetloss. My friend was an admin at a datacenter where they had dual DS3 connections to AT&T. He got a 200ms ping to Hawaii and some nasty jitter. I told him I got a 140ms ping and virtually no jitter. He didn't believe me until I got some samples. Even Japan is only 160ms.
Now what were you saying about a "fact"?
The first two generations of stars we HUGE and died quickly, destroying any planets near them. That's the whole issue. Our planet is part of the first generation of planets that would last long enough for intelligent life to evolve.