Bell's Own Data Exposes P2P As a Red Herring 261
dougplanet writes with news from the Canadian-throttling front: "As ordered by the CRTC, Bell has released (some) of its data on how torrents and P2P in general are affecting its network. Even though there's not much data to go on, it's pretty clear that P2P isn't the crushing concern. Over the two-month period prior to their throttling, they had congestion on a whopping 2.6 and 5.2 per cent of their network links. They don't even explain whether this is a range of sustained congestion, or peaks amongst valleys."
Re:Glad to hear this. (Score:5, Informative)
It was so obvious, we know ISP's are the worst kinds of businesses, they oversell the bandwidth massively on the customer end and yet their backbones are pretty hardly ever used so they just end up cheating the consumer. It's basically extortion.
Article is pure shit (Score:5, Informative)
http://www.cbc.ca/technology/story/2008/06/25/tech-caip.html [www.cbc.ca]
Re:Glad to hear this. (Score:3, Informative)
Re:I knew it!! (Score:2, Informative)
If you're going to Canada you'll need to get abreast of the proper Canadian dialect so you're able to voice your frustrations properly. So instead of saying "I'm coming down to your HQ and throwing a cinderblock through your front window!" it would be "I'm going to come down to your HQ eh, and I am going to throw a brick eh, through your window, eh!"
In all seriousness though. I hope this ruling will help in the fight against the plans to start charging for a monthly bandwidth allocation that Time Warner is setting up in response to 'congestion'. If you go over Time Warner's allocation they will begin charging you X dollars per MB over your allotment. I swear I went to high-speed internet (DSL) to get away from pay-as-you-go service. It's like when AOL 2.5 was around where you had to pay per minute, and finally they realized they'd get more business with 20$ a month unlimited minutes. That was the happiest day of my life, or at least it was until we got RoadRunner from Time Warner.
load of BS (Score:5, Informative)
In revealing the details, Bell explained in an accompanying letter that "while these numbers may seem low to the average lay person, they are significant to network traffic engineers such that it is important to consider the number of congested links in the proper context." - of-course, the context being that Bell would like to make more money from various throttling schemes as well as from their new IPTV stores.
If only a single link in the network is congested, end users may still experience slowdowns or dropped connections, the company said, - of-course, especially if you throttle these connections.
because the situation is similar to the road system -- where if one major artery is backed up, all connected roads will also have problems. - of-course they conveniently omit the fact that the Internet is designed to route around damaged/congested areas.
Re:How funny (Score:5, Informative)
Delays occur whenever anything is waiting in an output queue instead of being immediately transmitted. This could happen at very low average utilization levels if multiple sources all try to send data across a link simultaneously [formortals.com]. The delay time is a function of the number of bytes waiting to be transmitted and the transmit speed.
Retransmission delays occur when the output queue gets full, the router drops additional packets as they come in, and the TCP connection hangs until the retried packets come through (700ms for the first one, much more for subsequent dropped packets). To avoid compounding the problem, output queues on routers are typically sized to something a fair bit less than 700ms.
Interested Party Responses. (Score:3, Informative)
More interesting than the bell data are the responses from the other concerned parties.
Specifically, the response from Skype is a good read. The response from Cisco is pure crap and doesn't directly address the issue at hand.
Anyways, if you want to see the data yourself, look at the links here.
http://www.crtc.gc.ca/PartVII/eng/2008/8622/c51_200805153.htm [crtc.gc.ca]
Bell zip file with data: .doc. Go figure.
http://www.crtc.gc.ca/public/partvii/2008/8622/c51_200805153_1/920764.zip [crtc.gc.ca]
Note that all the Bell responses are in
Skype response:
http://www.crtc.gc.ca/public/partvii/2008/8622/c51_200805153/920240.PDF [crtc.gc.ca]
Cisco BS:
http://www.crtc.gc.ca/public/partvii/2008/8622/c51_200805153/920258.PDF [crtc.gc.ca]
Re:Glad to hear this. (Score:4, Informative)
Bank needs to repay you. (Score:5, Informative)
The bank needs to repay you the money they let Bell steal.
If a bank allows money to be withdrawn from an account of a deceased person, then the bank is liable to put the money back WITH interest and penal charges.
Once a person dies, the bank needs to legally freeze the account to prevent any deposits or withdrawals (esp. withdrawals).
Only the estate or the nominee can withdraw (not deposit) ALL the money from the account in one single operation.
Nope, the bank cannot unilaterally close and send you a check for the same. If you are the legal heir, you need to either prove by way of nomination OR successor OR court orders asking the bank to pay you the money.
The check that the bank cashed and the Direct Debit, if both happened AFTER your dad died, are not valid. In a court you WILL prevail, plus the bank has to pay a nasty fine.
But BEll cannot be held liable. You were not in a contractual relationship with Bell.
Order the bank in writing stating facts and giving them 7 days to repay you with interest.
If the bank fails to respond, file a criminal case stating fraud, and simulatenously ask the court to rule in your favor citing your dad's death certificate and date of debit.
The court usually will not want to hear from the bank because if the debit happened AFTER death then any legal arguments are moot.
Get a court order making the bank pay you.
If you want to play real nasty, send the order by ordinary post undistinguishable from other letters (after all banks hide their rate increases in same way) to the bank's registered office (NOT the branch). Those morons at the registered office will have no clue and throw away the letter. (Assuming you have given a deadline to pay you from date of letter do next steps).
Approach the court again whining pitifuly (yes it pays) that the Holy Judge's order was disobeyed (get the same judge) by an unruly bank.
The judge will ask what you want to do next.
This is most important: Now the culpability of the bank is established as defying court orders (your money now plays a second role. Judges don't like to see anyone defying their orders). Request the court grants you permission to seize and auction the bank's nearest branch's assets to get your money back. The judge will accept this.
Go with a sheriff and his posse to the branch, and now you are legally authorised to rob the bank. You can shut down the doors, throw out customers, restrain staff, seize cash from tills, auction PCs on the spot (better yet, arrange a few friends to be there for the auction to get bank's PCs at HUGE discounts). Sell ALL their stuff to get your money back: Remember, your goal is to first bankrupt the branch. Don't seize cash. Seize the hardware, valuable furniture anything that the bank needs to run its branch. Sell it on doorfront with sheriff standing by for a dollar or whatever you like.
The bank will try to move mountains to get the order overturned. So do it quickly, very fast. Get some 100 friends to suddenly appear, bid for the assets, and block the traffic to prevent their lawyers from reaching you to serve you a STOP SALE order they can get from a sympathetic judge.
Good luck
Re:Glad to hear this. (Score:1, Informative)
I'd switch in a fucking instant, right now, give me an alternative, I'm in.
Whats that? Oh thats right, there isn't one. Theres no competitor I can run to because everyone who supposedly 'competes' with bell is actually renting their lines anyway, and bell has demonstrated their willingness to fuck with third parties leasing their shit.
I could go for cable, you say? Nope same deal, the cable companies set up little pocket monopolies, if your in this city is rogers, one city over somebody else, never a choice. And all their packages are surprisingly shitty anyway.
Telcos long ago figured out they stood to make way more money for providing shittier service if they all provided the same shitty service.
And until somebody decides to spend a few trillion dollars to lay their own wires we'll never see competition. As long as the actual medium is controlled by one corporation (or a group working together) the status quo will never change, because a change would mean less money to The Suits.
Re:How funny (Score:5, Informative)
Kristof Obermann, he used to work at Arcor one of the bigger ISPs and phone companies in germany. He now works as a professor.
Besides, it's not like twice as much bandwidth costs twice as much money. Besides there's always redundancy. So in case one line breaks, you'll be at 100% peak utilisation. And then you will have problems as you will loose packets.
If those situations would exist, ISPs would use the TOS fields. Packets belonging would just be dropped more likely in case of a network overload. Nobody notices a missing packet at a download as the server will continue sending packets and the missing one eventually gets retransmitted a few seconds later. Missing packets are a _lot_ more noticable at web-browsing or interactive sessions.
Re:Harm done. (Score:5, Informative)
You're completely incorrect because you're completely ignoring Bell's criteria for a line being considered congested.
First, their thresholds: They consider anything over the following utilization thresholds to be congested:
DS-3 61%, OC-3 84%, OC-12 and OC-48 90%.
Second, they determine usage and congestion by taking samples every 15 minutes. If five samples return percentages over those limits in a 14 day period, the line is considered to be congested for the entire 14 day period.
Their percentages are actually not all that bad; they're a useful guideline for when it's time to turn on another link. Their methodology for MEASURING the usage, on the other hand, is completely flawed. A two-hour long DDoS attack one afternoon might mark a slew of lines as congested for two entire weeks.
Further bolstering the fact that they've chosen their measurements to make the issue seem worse than it appears is that despite the supposed congestion on a given percentage of their lines, they only have about 4000 ATM cell loss events network-wide each month. This is out of the trillions of ATM cells flying around their network every month, they only drop a percentage so small that my calculator resorts to scientific notation trying to calculate it.
In short, they've pretty much made up the issue. Their figures when taken at face value don't indicate significant congestion (5% of lines congested? Why not just purchased a handful more lines?), when examined based on their methodology appear to be garbage data, and when compared against actual packetloss caused by congestion, is proven to be completely non-existent. Bell has zero actual network congestion, their own ATM loss data backs that up.
Disclaimer: I'm not a network engineer, and so I might be talking out of my ass. But I think that common sense can play a role here; their methodology makes it trivial to declare a line as congested, and having 4000 instances of ATM cell loss on a network in a month with millions of customers (and trillions of ATM cells sent per month) doesn't seem particularly bad.
Re:Glad to hear this. (Score:2, Informative)
Sorry I have to say this: I feel like puking at the thought that /.ers actually find the above troll statement as informative. This is where public moderation fails in a horrible way: this is a popularity content, not moderation
Regarding the original issue, I worked for a few years as an admin on an ISP and I can tell you torrents have some nasty impact on the routers. Bell just really sucks on presenting the issue. Torrents are basically DOS, and deep down you know it. All ISPs are hating it, and not because of the copyright issue (they don't care about WTF you steal), its because of technical reasons(they care about you smashing their routers/lines).
Re:Harm done. (Score:2, Informative)
/QFT. I was thinking the same thing when I read the op.
Congestion on only a few end links can cause a whole interlink to get saturated. I.E. if you have 128 customers on a DSLAM box with 10Mbps to the backbone, and 8 of them saturate their 1.5Mbps link, that's all 128 dealing with shitty network throughput since they connect up to a core link through the same wire. If you multiply that out, the whole backbone gets saturated and everyone gets horrible service.
That's *less* than 5% of people saturating their connections. If they are doing it 24x7 they become a nuisance and the other 123 people can't even look at cnn without experiencing stalled page loads.
P2P _must_ be throttled or no one's QoS can be guaranteed and they'll lose customers. What they should really do, is scale p2p based on demand... IE if someone is saturating their connection as others start sending packets, back off their connection speed as necessary. They shouldn't target p2p traffic specifically, they should apply it to _all_ traffic. Otherwise p2p clients will just start operating on random ports and their rulesets will get out of hand, slowing things down even more.
Trust me, I hate comcast and verizon as much as the next guy, but they gotta do what they gotta do. You can't bend the laws of physics. People that whine about it have sharing issues and are selfish.
They should simply meter it and charge people for what they use beyond 50GB a month or something. "You want to p2p constantly? Fine, we need to upgrade our network to accommodate you, so it's going to cost you."
This would serve fine for people that occasionally need to download a 8GB game they just bought from direct2drive or something, but p2p would get real expensive, real fast, for people that simply log into whatever p2p system and download everything they see.
-Viz
Re:How funny (Score:3, Informative)
I don't know if this is related to the GP's post, but a common reason to cap utilization at 50% (or upgrade the link when you approach 50%) is because of redundancy: given two separate links each carrying half the load and where the links are not used beyond 50%, if one link fails, the other can pick up the remaining load without significant (perhaps slight) service degradation.
Re:And in other news... (Score:2, Informative)
While the cap may be 30GB, quite a large % of ISP's still shape P2P traffic. You will find that the majority of the large providers do. That being said I generally get fantastic speeds using ftp. 20$ a month for fiber is a great deal here... To bad about the N.A market.
Re:Harm done. (Score:3, Informative)
okay, it's simple : a datapipe is not a water pipe. It's serial, not parallel.
In a water main every droplet of water has it's own space available, in an OC-48 every bit must be nicely lined up and 1 bit at a time gets sent down the line. (yes I know about the modulation, so it's 64 bits or so, but the principle is the same)
That means that a line that is 90% utilized in a space of 15 minutes is 13.80 minutes utilized and 1.20 minutes not utilized at all : there could be a single 13.80 minute burst and then nothing and the line would be 90% utilized. Generally you're going to see a limited number of bursts per minute.
The bursts are unpredictable, but they get exponentially longer the higher the utilization gets. A 1% used line will in practice only have bursts of milliseconds. In a 90% used line the bursts will (sometimes, obviously) last minutes.
During a burst, basically the pipe is utterly full. Nothing can get in until enough stuff gets out first.
Therefore a 90% utilized line is totally unuseable for anyone. It leads to massive slowdowns.
I know this is not what you want to hear, but it's the way it is, simple as that.