en102
Canadian, eh?
join:2001-01-26
Valencia, CA | reply to dadkins
Re: Waiting...
It would be easier to deal with overloaded nodes at times than going through caps/filters/traffic shaping techniques.
--
Canada = Hollywood North |
|
TKJunkMail
Enjoy the sun
Premium
join:2002-03-03
Avalon, NJ
 ·Sprint Mobile Broa..
 ·Comcast
| said by en102  :It would be easier to deal with overloaded nodes at times than going through caps/filters/traffic shaping techniques. Only for those CAUSING the overload. For everyone else, this will be an improvement.
--
My BLOG .. .. Internet News .. .. My Web Page |
|
en102
Canadian, eh?
join:2001-01-26
Valencia, CA
·RoadRunner Cable
 ·DSL EXTREME
| I don't know.. I've never had issues with an overloaded node on cable or DSL... which is why I see this as a method of rate increase for consumers.
Eg. Lowest tiers ( 1.5Mbps or less) should not need caps, yet they're typically capped hard to purchase a higher rate plan, which is now set as an overloaded node, and requires traffic shaping and/or caps. The end result for the AVERAGE user is not much (i.e. going from a 1.5 to 10Mbps) unless they are bit torrent / gamers / VOD users.
--
Canada = Hollywood North |
|
nasadude
join:2001-10-05
Rockville, MD
·Comcast
| reply to en102
said by en102 :It would be easier to deal with overloaded nodes at times than going through caps/filters/traffic shaping techniques. everything I've read indicates the most straightforward, simplest way to deal with congestion is to add capacity. One of the network engineers for Internet2 said they did a trade between caps/filters/shaping/etc. and adding capacity and adding capacity was cheaper and simpler.
I would be willing to bet money the real reason comcast is doing this is to position themselves for the "two tier" internet, with "fast lanes" for their content and those that pay a premium and "slow lanes" for the rest. |
|
en102
Canadian, eh?
join:2001-01-26
Valencia, CA
·RoadRunner Cable
·DSL EXTREME
| I'm sure that adding capacity is simpler, and very effective.
ISP's (management) typically would prefer making more money off of existing, through attempting to squeeze a lump of coal into a diamond. Management doesn't really 'care' if its simpler, your typical upper level management is type alpha, and would prefer control at all cost.
I.e. Simple solution = add capacity (requires cost to an outside source... no direct gain, capacity will hit again in the not too distant future)
Complex solution = form of traffic reduction through many different means, which will require some cost, but can recoup at the per MB transferred level, and will allow for future transparent price increases (i.e. same price for service, higher cost on overage, etc)
--
Canada = Hollywood North |
|
espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
 ·voip.ms
·Vitelity VOIP
·Callcentric
·VoiceStick
·ViaTalk
·Comcast
·Embarq
| reply to en102
said by en102 :I don't know.. I've never had issues with an overloaded node on cable or DSL... which is why I see this as a method of rate increase for consumers. Unfortunately your experience doesn't necessarily match that of all other Internet users. Just look at the poor bastards on Embarq service in Indiana that regularly see latency shoot to 300+ms round trip because Embarq doesn't throttle anyone:
»Embarq network congestion, latency and packet loss in IN
P2P users don't care - they can fire up their client and walk away. For anyone else trying to use VoIP, game, or even just surf the web the situation just sucks. |
|
Matt
Take me down to the paradise city
Premium
join:2003-07-20
Jamestown, NC
·North State Commun..
| reply to nasadude
said by nasadude :I would be willing to bet money the real reason comcast is doing this is to position themselves for the "two tier" internet, with "fast lanes" for their content and those that pay a premium and "slow lanes" for the rest. Yep, and Comcast's pursuit of P4P (Bittorrent that prefers local, on-network peers) is evidence of that. If they really were concerned about local node congestion, there are MUCH easier and more effective ways to handle this. This is nothing more than a way to kill P2P video distribution and protect their Video On-Demand revenue, while simultaneously lowering costs.
Time Warner is doing the same thing, only with a ridiculously low bit cap. |
|
en102
Canadian, eh?
join:2001-01-26
Valencia, CA | reply to espaeth
True - not all nodes are good on all providers.
DSL-Extreme has been good.
--
Canada = Hollywood North |
|
espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
·voip.ms
·Vitelity VOIP
·Callcentric
·VoiceStick
·ViaTalk
·Comcast
·Embarq
1 edit | reply to nasadude
said by nasadude :everything I've read indicates the most straightforward, simplest way to deal with congestion is to add capacity. One of the network engineers for Internet2 said they did a trade between caps/filters/shaping/etc. and adding capacity and adding capacity was cheaper and simpler. When technologies like Ethernet and Packet-over-SoNET are being used, you are correct it is usually cheaper to add bandwidth. With technologies like DSL and DOCSIS, there are limits to how far you can expand them out. In the Internet2 domain they can add a variety of links from 100mbps - 10,000mbps (10gbps). For DSL you're limited by distance, protocol, and collective connection crosstalk. (ie, if all lines are transmitting at the same time the effective throughput goes down on each line due to interference) For DOCSIS you can only increase downstream capacity 38mbps at a time eating up 6MHz chunks. Even there, the DOCSIS standard limits how many channels you can have on a segment.
The ugly truth is that broadband networks have much harsher scaling limits than other network technologies, and upgrades often require forklift replacement of all involved distribution gear. Just look at DOCSIS 3.0 -- MSOs won't see the benefits there until all of the head-ends and most user cable modems are replaced. Swapping out 14+ million end-user devices isn't a swift activity. |
|
ske99slem
@swbell.net
| reply to nasadude
said by nasadude :said by en102 :It would be easier to deal with overloaded nodes at times than going through caps/filters/traffic shaping techniques. everything I've read indicates the most straightforward, simplest way to deal with congestion is to add capacity. One of the network engineers for Internet2 said they did a trade between caps/filters/shaping/etc. and adding capacity and adding capacity was cheaper and simpler. I would be willing to bet money the real reason comcast is doing this is to position themselves for the "two tier" internet, with "fast lanes" for their content and those that pay a premium and "slow lanes" for the rest. Adding capacity is expensive and no one here seems to want to pay for it. If they can't get more revenue, why do you think they would pay for more network?
Adding a few shaping boxes is relatively cheap and solves the problem in the short term. In the long term, us high usage people are going to be paying for their capacity upgrades.
Doc |
|
dnoyeB
Ferrous Phallus
join:2000-10-09
Southfield, MI | reply to Matt
I agree on the VOD front. The writing is on the wall with the new streaming p2p direction. |
|
jc100
join:2002-04-10
| reply to espaeth
Simple solution. Limit the speed of high users, hence alleviating congestion and free up speed for everyone else. If everyone has 10/1 and say 10 percent of users exceed a CLEAR CUT TRACKABLE CAP (IE offer software or online counter to track usage), these 10 percent get throttled to a slower speed. Say maybe 3/512 or 2/512 for the remaining month. This A) Cuts down on lawsuits and investigations as no traffic shaping is taking place. B) ISPS say UP TO in their speed clause meaning they can offer less. C) Make it clear to users. This way you don't alienate everyone. Also, no one is disconnected. The worst that happens is their speed is slower. |
|
dynodb
Premium,VIP
join:2004-04-21
Minneapolis, MN
| reply to en102
said by en102 :It would be easier to deal with overloaded nodes at times than going through caps/filters/traffic shaping techniques. If it were just dozens of nodes- perhaps.
If it's thousands of nodes- probably not.
Besides, increase the bandwidth and the "bandwidth hogs" just use that much more. Double the capacity today, and a month from now they'd be in the same boat. |
|
espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
·voip.ms
·Vitelity VOIP
·Callcentric
·VoiceStick
·ViaTalk
·Comcast
·Embarq
| reply to jc100
said by jc100 :Simple solution. Limit the speed of high users, hence alleviating congestion and free up speed for everyone else. That's exactly what it sounds like Comcast is implementing here, no? |
|
jc100
join:2002-04-10 | Nope they are instituting usage caps (X amount allowed) versus speed caps (no limits on usage but limits on bandwidth capacity). I think a bandwidth capacity limit would be better than a usage cap. My 2 cents. |
|
espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
·voip.ms
·Vitelity VOIP
·Callcentric
·VoiceStick
·ViaTalk
·Comcast
·Embarq
| Actually, I don't disagree with that at all. The key constraint in networks is rate, not quantity, so it only makes sense to come up with a good mechanism for balancing the traffic rates that each subscriber gets. In the hosting world quantity based limits are loved by users and hated by providers, because all too often someone will wait until the last 3 or 4 days of the billing cycle to decide they want to move 400GB of data but still be under their 1500GB/mo limit.
The problem is that most users have trouble with the difference between rate and quantity when it comes to talking about bandwidth. It happens in nearly every thread on this site.
I wonder if there is any good way to bridge that gap in understanding to arrive at a shared network plan that is truly fair? |
|
en102
Canadian, eh?
join:2001-01-26
Valencia, CA
·RoadRunner Cable
·DSL EXTREME
| reply to dynodb
I tend to agree that there's a limit at both ends.
1. Don't sell something (i.e. +20Mbps/1Mbps) then cap/filter because you can't reasonably sell that product anyways. Eg. If everyone was at 1.5Mbps, there would probably be no issue on that same node.
2. 20Mbps 'looks' better for sales than '1.5Mbps', especially for video apps. E.g. You 'need' to have this package to do 'X'.
Its all marketing, and a way to ensure that the 'unlimited' isn't 'SYN FLOOD'.
--
Canada = Hollywood North |
|
jc100
join:2002-04-10
| reply to espaeth
Well the honest truth is this, and honestly this. You have a large share of users who are not tech savvy. They are susceptible to hackers, viruses, botnets, etc. These users wouldn't know the difference if they used the bandwidth or someone else. The other half are tech savvy but probably could care less. You have a small amount who given the tools would be conscience of their usage. Therefore, my solution above is the most reasonable. For the less tech savvy, their bandwidth is slowed down. It will cut usage on the network, and make them solve the problem of why they consume bandwidth (virus, hacked, etc). For the other portion that like to download a ton, capping them with speed limits would make them more conscientious. Let's face it, people don't want to wait a long time on their download. Making a 10mbit line 2-3mbit down is quite a punishment. It is enough to where it might discourage high users but not to the point where it pisses them off. Plus, it lessons the strain on the network. I know some foreign isps do this. Cap high users during peak hours. It would be a start. |
|
funchords
Hello
Premium,MVM
join:2001-03-11
Washington, DC
·Verizon Online DSL
·Skype
1 edit | reply to espaeth
said by espaeth :The problem is that most users have trouble with the difference between rate and quantity when it comes to talking about bandwidth. It happens in nearly every thread on this site. Q. What's the difference between 60 miles per hour or a mile per minute?
A. None
Bandwidth is bandwidth is bandwidth. If you have an amount divided by a time period, that's an expression of bandwidth.
Only in Comcast land do people invent concepts to purposefully cloud the issue. What's surprising is that people actually buy it and repeat it!
--
Robb Topolski -= funchords.com =- Hillsboro, Oregon
HTTP is the new Bandwidth Hog...
|
|
espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
·voip.ms
·Vitelity VOIP
·Callcentric
·VoiceStick
·ViaTalk
·Comcast
·Embarq
| said by funchords :Q. What's the difference between 60 miles per hour or a mile per minute? 60 miles per hour could be 50 minutes of 0 miles per minute, and 10 minutes of 6 miles per minute.
said by funchords :Bandwidth is bandwidth is bandwidth. If you have an amount divided by a time period, that's an expression of bandwidth. But interfaces are only capable of moving so many bit per second. It doesn't matter if 1mbps is roughly 330GB/mo -- that doesn't mean you can try to move 330GB in just one of those 30 days and have it work.
This is like reporting how many cars a highway can move per hour vs how many cars it can move per day. Those are vastly differently numbers with very different meanings.
You're demonstrating my point quite nicely though, Robb. You're an incredibly sharp guy (I mean this genuinely, no BS), and yet you're falling into the same pitfalls most on this site do. |
|
