 espaethDigital PlumberPremium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2
 ·Clear Wireless
2 edits | reply to r81984
Re: In one sentence. said by r81984:I am not talking about ARP requests or broadcast traffic. It seems like you know what I am talking about but just want to pretend they can accurately count traffic. You don't have to feed me BS, I have first hand experience trying to accurately record bandwidth from the switch level and there is no way with TCP/IP to make it accurate. Clearly you have to know what you're doing, namely make sure you're using 64 bit counters vs the standard 32 bit counters that many try to use. Still, this is a problem the industry as a whole solved over a decade ago. Level(3), Cogent, Sprint, MCI, ATT, Global Crossing, PCCW/BTN, AboveNet, Hurricane Electric, Internap, Savvis, Telia Sonera, Peer1, and pretty much every other Internet carrier out there has successfully cracked this nut.
said by r81984:Also, everyone is vunerable from virus and spyware traffic that they cannot control. Having your computer infected for 24 hours could give you huge overages and costs. Again you can't trust any of these ISPs with case by case rule. First off, point out a single residential broadband proposal where someone could get a $10k bill for metered broadband. Even the Time Warner proposal that caught so much heat had a maximum cap of $75 in overage charges. So if you were on the $75 top tier plan, the worst monthly bill you could ever see is $150. It's incredibly easy to mitigate these types of issues from becoming financially crippling by implementing simple policies like overage caps.
said by r81984:If you went to pay by the byte right now thousands would get giantic bills at the end of the month just from viruses and spyware. If you ever had to work internet tech support you would know this. The publicly available statistics don't support that, such as Cisco's broadband traffic study or several of the reports available at »www.dtc.umn.edu/mints/home.php
The other problem with this assertion is that ISPs already have verbiage in their ToS today that allows them to evict heavy users from the network due to either openly or not-so-openly disclosed usage limits. The folks who would be facing your proposed multi-thousand dollar overage would have already been ejected from the network due to ToS usage violations.
"Unlimited" usage is a farce. ISPs either took their own swag at usage, or actually invested in hiring some bean counters with actuarial sciences knowledge to price their service offerings. Every company has a magic value of bandwidth that they want to sell you for your monthly subscription rate. Capacity can always be augmented for more money, so they put a price on usage based on the estimated costs of future expansion. Everyone who uses less than that amount of bandwidth is profitable, everyone over that amount is unprofitable. Reach a certain level over the unprofitable mark, face ejection from the network.
That's the biggest risk to the current flat-rate system of billing. It only works so long as subscribers conform to the ISP's planned amount of usage.
said by r81984:No one is forcing ISPs to give 20mpbs, if they up their speeds without increasing the bandwidth to their network then oh well, not our problem. If they are overselling the network everyones connection rate should go down, but that never happens. Why don't they upgrade their networks with the extra money? The link between quantity usage and provisioned rate is quite possibly the most misunderstood / misrepresented concept in networking.
Network access is essentially random, which leads favorably to a statistically muxed or oversubscribed configuration. The key variable in all of this is time. Giving users higher provisioned speeds actually has what many would consider to be a counter-intuitive effect: it makes even greater levels of oversubscription work better. The reason behind this is simple: you're going to use the network to get the content you want, regardless of the speed. The faster you can get the content, the less time your traffic will be present on the network, and the less likely it is that your file transfer will coincide with other user's file transfers on the network.
To say that ISPs haven't been upgrading their networks is a huge misstatement. They absolutely are -- that's why you're seeing DOCSIS 3.0 rollouts, Qwest & ATT doing deeper fiber deployments, Verizon doing FTTH deployments. The problem with flat-rate pricing is that upgrades are also flat-rate budgeted. Upgrades come as money is pooled to support the infrastructure increases, which have absolutely no link to demand other than being a selection criteria for where upgrades will be deployed first. |
|
 r81984Fair and BalancedPremium
join:2001-11-14
Katy, TX
 ·AT&T U-Verse
 ·AT&T DSL Service
·row44
1 edit | "64 bit counters vs the standard 32 bit counters"
LOL. You have no idea what that means.
The different between those two do not make it more accurate for legit and non legit traffic. It is amazing what people try to pretend they know these days. Just Wow. You just told me that you have no idea how stats are collected.
If they charge $1 per Gigabyte and someone easily sends garbage traffic (that looks legit, very easy to do) to your IP then you will have a large bill.
If your computer gets a virus and taken over then you could get a large bill.
Also as reported on another topic netflix streaming is at 1GB an hour and even more for HD. Hopefully you don't have netflix.
If they truely are hurting for money they need to raise the monthly connection fees, not try to screw people and limiting the internet by charging by the byte. Charging by the byte will not benefit anyone and as user average usage increases due to higher bandwidth websites we will all be screwed.
Comcast is not hurting with flat rate billing: »news.yahoo.com/s/nm/20091104/med···_comcast
If ISPs ever go to metered broadband I feel sorry for the family of 6 with 4 teenagers.
--
Democrats are not Socialists any more than Republicans are. |
|
espaethDigital PlumberPremium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2 Reviews:
·Clear Wireless
| said by r81984:"64 bit counters vs the standard 32 bit counters" LOL. You have no idea what that means.The different between those two do not make it more accurate for legit and non legit traffic. I never said it made it more accurate for "legit and non-legit" traffic. The places where many people run into issues with SNMP polling of interfaces is if the interface counters wrap too quickly in between polling cycles, which is more likely to occur with 32bit number space.
As for "legit vs non-legit" traffic, that's a non-factor. "Non-legit" traffic, as you define it, is the statistical exception, not the statistical norm. Yes, people could definitely generate tons of traffic addressed at an end-user that isn't part of an active or established connection ... yet in most cases that doesn't happen. In fact, you are statistically more likely to get in a fatal automobile accident than have that occur -- are you going to quit driving out of fear?
Being in a support position of course you are going to see this occur with a greater amount of visiblity. Same deal with EMTs who spend their days cleaning up the after effects of automobile accidents...
said by r81984:If they truely are hurting for money they need to raise the monthly connection fees, not try to screw people and limiting the internet by charging by the byte. Ok. So if 1% of your subscriber base is using resources at a level that causes you to incur further financial impact, you're saying 100% of the subscriber base should have their rates raised because of it?
said by r81984:Charging by the byte will not benefit anyone and as user average usage increases due to higher bandwidth websites we will all be screwed. The same way pre-paid cellular options screwed over people?
There are cases where usage-based billing can make sense, and can be effective.
Comcast has a 250GB residential usage cap. You use too much bandwidth, you get a call. Go over the limit again, you get punted from the service.
Flat-rate isn't unlimited. You keep pointing out cases where usage will increase, but in the case of flat-rate billing you don't get the option to pay more -- you just get the option to conform or be kicked out. |
|
|
|
r81984Fair and BalancedPremium
join:2001-11-14
Katy, TX Reviews:
·AT&T U-Verse
·AT&T DSL Service
·row44
1 edit | Just imagine if 1/2% of all electricity meters and water meters in this country were wrong and there was no way to prove they were wrong.
There will never be a 100% way to monitor bandwidth unless they install software on each end users device.
Also, since the costs of the network are fixed based on each connection there is no reason to come up with some BS variable billing because if done properly the company would end up loosing money as it was previously posted that there are way more people below average than above. They will only implement it in a way that hurts consumers and makes them more money.
If they want more money they just need to raise the flat fee instead of punishing people who untilize their connection.
250GB cap is way too low, with netflix that could be less than two movies a day. Again hopefully you dont like to use itunes, netflix, hulu, xm online, xbox live, or watch online news, youtube, upload videos, etc.
Anytime you have have a cap you are unfairly restricting the internet and limiting online businesses. For the internet to work it has to be free and unrestricted.
I am just curious which monster ISP do you work for?
It is obvious your views are not based on reason or experience, but on a job or something else.
--
Democrats are not Socialists any more than Republicans are. |
|
espaethDigital PlumberPremium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2 Reviews:
·Clear Wireless
| said by r81984:Just imagine if 1/2% of all electricity meters and water meters in this country were wrong and there was no way to prove they were wrong. How do you know that's not the case right now? How many people have a personal meter besides the utility electric or water meter? The only option most people have to validate their utility bills is to "gut check" the bill against previous billing cycles.
In the case of broadband you can measure bits in & bits out on your router and compare them to the ISP values. This isn't rocket science, despite your best efforts in trying to make this more difficult than it needs to be.
said by r81984:Also, since the costs of the network are fixed based on each connection there is no reason to come up with some BS variable billing because if done properly the company would end up loosing money as it was previously posted that there are way more people below average than above. They will only implement it in a way that hurts consumers and makes them more money. Metered billing isn't about "$0 base + usage" billing; to sell this as "your bill will go down" would be disingenuous. This is really about curbing how much broadband bills will go up from this point forward. Any system will most likely be a bucket + overage system, the same as cell phones with plan minutes.
said by r81984:250GB cap is way too low, with netflix that could be less than two movies a day. Again hopefully you dont like to use itunes, netflix, hulu, xm online, xbox live, or watch online news, youtube, upload videos, etc. Not everybody uses these services to excess, and there are actual statistics that show that.
There are approximately 228 million Internet subscribers in the US, of which approximately 70 million are broadband subscribers. (source: »www.internetworldstats.com/am/us.htm )
According an October 2009 report, 168 million unique viewers watched online videos of some kind. Hulu averages 38.7 million unique viewers a month viewing an average of 15.1 videos. So about 17% of all US Internet subscribers are hitting up Hulu and watching an average of 15.1 videos. So that's like 30 hours of video, right? Not according to the statistics...
•The average Hulu viewer watched 15.1 videos, totaling 1 hour and 32 minutes of videos per viewer.
•The duration of the average online video was 3.8 minutes.
Source: »www.comscore.com/Press_Events/Pr···ewership
The average broadband subscriber uses somewhere in the neighborhood of 11GB/mo.
Sources: »newsroom.cisco.com/dlls/2009/pro···109.html
»www.dtc.umn.edu/mints/home.php
This is the kind of usage that network planners are building for, and the kind of usage that the flat-rate billing model is based upon. Your assertion that you should be able to dramatically increase average usage without any impact to cost is overly idealistic, I'm afraid. The cold evil bitch that is reality always wins.
said by r81984:I am just curious which monster ISP do you work for? I'm a network architect for a Fortune 100 Healthcare organization. I have every interest for broadband service to be as cost effective as possible, as we have made significant reductions to overhead by dismantling call centers and turning agents into work-at-home employees. It would be great if the ideas you presented were true and that increasing utilization actually had no impact on cost, but sadly that's just not the case. |
|
| reply to r81984
said by r81984:Just imagine if 1/2% of all electricity meters and water meters in this country were wrong and there was no way to prove they were wrong. There will never be a 100% way to monitor bandwidth unless they install software on each end users device. Dude, you have no idea what you're talking about. "legit" vs. "non legit" traffic? Please. It came down your line, so you need to pay for it. If your water line develops a leak after the meter then you should not have to pay your water bill because you did not "use" it either, right? How about if your portable electric heater malfunctions and turns itself on in the middle of summer, causing your A/C to work harder? I guess you think the utility company should eat that cost too.
If you infect your computer and it starts sending 10's of gigs of SPAM, then FIX IT! If metered billing does go into effect, then it's now in the end users' financial best interest to get their machines cleaned. If you need to pay someone to do that, tough luck.
Fight against metered billing with FACTS, not a bunch of whining about how you do not want to be responsible for your own machine.
/mackey |
|
r81984Fair and BalancedPremium
join:2001-11-14
Katy, TX Reviews:
·AT&T U-Verse
·AT&T DSL Service
·row44
| With water and eletricity you know what you appliances and fixtures use in advance.
With a computer you have no idea in advanced how big a streaming video is or how big a website is.
"FIX IT"???? Because 90% of the population is computer experts, just read espaeth posts.
If ISPs want to reduce bandwidth then they should fix peoples computers. Also your computer can be turned off and garbage data can still be sent to your IP, the ISP switch or modem will still count that as data. There is no way for them to know what is legit and what is not.
--
Democrats are not Socialists any more than Republicans are. |
|
r81984Fair and BalancedPremium
join:2001-11-14
Katy, TX Reviews:
·AT&T U-Verse
·AT&T DSL Service
·row44
| reply to espaeth
said by espaeth:I'm a network architect for a Fortune 100 Healthcare organization. Good one.
I guess your the guy that plugs in the equipment not the guy that sets it up?
ISPs need to stop increasing their speeds if their networks cannot handle it. I think comcast had it right when they had lower speeds, but used speedboost. They are just wrong about using caps.
It is silly to call it 20mbps and then limit you to 250 GB a month.
Techinically 250 GB a month is only 788 kbps.
To be truthful with a 250 GB cap they should only be advertising average speed 788 kbps with speedboost up to 20 mpbs.
--
Democrats are not Socialists any more than Republicans are. |
|
| reply to r81984
said by r81984:With water and electricity you know what you appliances and fixtures use in advance. With a computer you have no idea in advanced how big a streaming video is or how big a website is. "FIX IT"???? Because 90% of the population is computer experts, just read espaeth posts. If ISPs want to reduce bandwidth then they should fix peoples computers. Also your computer can be turned off and garbage data can still be sent to your IP, the ISP switch or modem will still count that as data. There is no way for them to know what is legit and what is not. One thing you managed to accomplish in this thread is to document that you have zero understanding of how accurate IP accounting is actually.
Not only is it possible to count your "legitimate" and "non-legitimate" traffic, it is possible to differentiate between them. So that argument is specious at best. The fact that you do not understand how that can be done does not make your assertion correct.
The very idea that ISPs should fix peoples computers is ludicrous. Next you will want the gas station to fix your because they sell you gasoline. |
|
1 edit | reply to r81984
said by r81984:With water and eletricity you know what you appliances and fixtures use in advance. With a computer you have no idea in advanced how big a streaming video is or how big a website is. I like how you completely ignored my leaky-pipe analogy and instead made a 100% COMPLETELY FALSE statement. Seriously, name how much water (or how many kWH) your dishwasher uses each cycle. What, you don't know? How about your washing machine? What I'm saying is you may not know *exactly* how much water your appliances use, but you have a general idea. Web pages are the same way: I may not know beforehand that a page uses 59.837 KiB, but I have a general idea due to past experience with similar sites. Ditto with movies: I know the movies I watch are about 200 MiB for every 25 min.
said by r81984:"FIX IT"???? Because 90% of the population is computer experts, just read espaeth posts. Seriously? I've read espaeths' posts, and the only wannabe computer expert here is YOU. I guess you just leave your car whenever it breaks down because you are not a mechanic. Like I said in my last post, If you need to pay someone else (because you are not a "computer expert") to fix it, then tough luck.
said by r81984:Also your computer can be turned off and garbage data can still be sent to your IP, the ISP switch or modem will still count that as data. Also 100% FALSE! When the guy/computer/device tries to send you "garbage data" and your machine is off, THE CONNECTION IS NEVER ESTABLISHED AND THUS NO DATA IS SENT. TCP/IP 101 dude. If you want to avoid the (~20 byte) connection attempt ("SYN") packet, then shut off your modem or router - this way your ISPs' router will not even know where (the MAC address) to send the connection request to. (please don't make me explain how ARP works...)
I'm assuming you expect cell phone companies to not charge you (use minutes) for telemarketer and wrong number calls, too?
/mackey |
|
r81984Fair and BalancedPremium
join:2001-11-14
Katy, TX Reviews:
·AT&T U-Verse
·AT&T DSL Service
·row44
| reply to AnonDOG
said by AnonDOG :
Not only is it possible to count your "legitimate" and "non-legitimate" traffic, it is possible to differentiate between them. So that argument is specious at best. The fact that you do not understand how that can be done does not make your assertion correct. I never said it was impossible, it just is impossible at the switch or modem level.
Unless they run software on each end device (PC, Xbox, etc) it cannot be accurate.
--
Democrats are not Socialists any more than Republicans are. |
|
r81984Fair and BalancedPremium
join:2001-11-14
Katy, TX Reviews:
·AT&T U-Verse
·AT&T DSL Service
·row44
| reply to mackey
said by mackey:I'm assuming you expect cell phone companies to not charge you (use minutes) for telemarketer and wrong number calls, too? /mackey You need to read up on TCP/IP 101.
With a cell you have caller ID. You do not need to answer the phone or you can just hang up and then you lose no more minutes. Your ISP has no idea which traffic is dropped by your computer.
--
Democrats are not Socialists any more than Republicans are. |
|
| said by r81984:You need to read up on TCP/IP 101. With a cell you have caller ID. You do not need to answer the phone or you can just hang up and then you lose no more minutes. Your ISP has no idea which traffic is dropped by your computer. Whatever dude, you're a moron.
And yes, your ISP *DOES* know, mainly because THAT TRAFFIC IS NEVER SENT OVER THE NETWORK if the connection is refused. In other words, NO DATA IS TRANSFERRED UNTIL YOUR COMPUTER ACCEPTS THE CONNECTION. If your computer is off then it cannot accept the connection and thus no data is sent to it. If you cannot grasp this concept then you have no business posting in this thread.
/mackey |
|
r81984Fair and BalancedPremium
join:2001-11-14
Katy, TX | "no data is transferred until your computer accepts the connection" LOL
It is even more funny when you through a home router into the equation.
--
Democrats are not Socialists any more than Republicans are. |
|
| ?? Just because you have absolutely no clue what you're talking about does not mean it's funny...
BTW, that's "throw" not "through"
/mackey |
|
r81984Fair and BalancedPremium
join:2001-11-14
Katy, TX Reviews:
·AT&T U-Verse
·AT&T DSL Service
·row44
| Metered billing at the switch or modem will never be 100% there will always be easy ways for someone to send you garbage data that will count against your usage.
ISPs need to setup their networks properly for the connections they have instead of limiting the internet.
--
Democrats are not Socialists any more than Republicans are. |
|
espaethDigital PlumberPremium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2 Reviews:
·Clear Wireless
| reply to r81984
said by r81984: Your ISP has no idea which traffic is dropped by your computer. This is not necessarily true.
Unidirectional traffic is very easy to spot using a variety of techniques. This can be full in-line packet inspection using anti-DoS appliances such as Tipping Point, Arbor Networks eSeries, or Cisco Guard appliances.
You don't need to necessarily place a device in-line. You can leverage the span / port mirroring capabilities on aggregation devices to send a copy of the packets to a server-based platform for further analysis and evaluation. Solutions like Arbor Networks Peakflow software, NetQoS, or nTop can all take a mirrored data flow and analyze the data to identify suspect traffic.
Any of these solutions can use heuristic analysis to look for specific patterns of data within packets, very similar to how a virus scanner identifies viruses.
Even still, you can identify the unidirectional flows you are talking about here by using software to analyze NetFlow data coming off any of the layer3 hardware on the ISP network to identify asymmetrical flows.
Here are some links to various products and implementations that do exactly what you are describing to be impossible:
»www.tippingpoint.com/technology_dos.html
»www.cisco.com/en/US/products/ps5···dex.html
»www.arbornetworks.com/
»www.flukenetworks.com/fnet/en-us···ucts.htm
»www.netqos.com/
»www.ntop.org/news.html
»en.wikipedia.org/wiki/Netflow |
|
