dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
7
share rss forum feed


fuziwuzi
Not born yesterday
Premium
join:2005-07-01
Atlanta, GA
reply to JohnInSJ

Re: Here we go again about bandwidth throttling ...

said by JohnInSJ:

said by fuziwuzi:

which in non-BizarroWorld is called "throttling"
Nope. In non-bizarroworld its called QoS
Sorry, but in my opinion your "QoS" is simply more double-speak for what is, effectively, throttling.
--
***************
I think computer viruses should count as life. I think it says something about human nature that the only form of life we have created so far is purely destructive. We've created life in our own image.
- Stephen Hawking


JohnInSJ
Premium
join:2003-09-22
Aptos, CA
said by fuziwuzi:

Sorry, but in my opinion your "QoS" is simply more double-speak for what is, effectively, throttling.
Sorry, but in Computer Science the actual correct term for assigning priority to packets based on the quality of service desired is called QoS.

Throttling is something you do to people who just cannot wrap their heads around a concept that there is only so much capacity in a system, and when you've saturated that capacity then some packets are not going to get through as fast.

The FACT that you cannot shove more packets through a pipe then it can actually handle means you have to SELECT some packets to go ahead of others to maintain overall Quality of Service. Simple as that.
--
My place : »www.schettino.us


fuziwuzi
Not born yesterday
Premium
join:2005-07-01
Atlanta, GA
said by JohnInSJ:

said by fuziwuzi:

Sorry, but in my opinion your "QoS" is simply more double-speak for what is, effectively, throttling.
Sorry, but in Computer Science the actual correct term for assigning priority to packets based on the quality of service desired is called QoS.

Throttling is something you do to people who just cannot wrap their heads around a concept that there is only so much capacity in a system, and when you've saturated that capacity then some packets are not going to get through as fast.

The FACT that you cannot shove more packets through a pipe then it can actually handle means you have to SELECT some packets to go ahead of others to maintain overall Quality of Service. Simple as that.
You get caught up in semantics. If I am transferring data and Comcast or anyone else at some point during that transfer causes that data transfer to be slower, by whatever method they choose, then effectively that data transfer has been throttled. You can choose to call it "QoS" or "packet prioritization" or a host of other terms, but effectively it is throttling. The data transfer becomes slower because of ACTIVE ISP manipulation. I am not talking about the passive capacity of "the pipe", which is irrelevant to the discussion, I am talking about active manipulation by Comcast of the effective (key word) speed.

Your ad hominem insults aren't necessary here.
--
***************
I think computer viruses should count as life. I think it says something about human nature that the only form of life we have created so far is purely destructive. We've created life in our own image.
- Stephen Hawking


JohnInSJ
Premium
join:2003-09-22
Aptos, CA
You're completely wrong however.

Packet prioritization only happens when the node is saturated. Which means SOMEONEs packets will not get through.

QoS marking of the packets that are causing the saturation means everyone gets a fair share of the capacity.

In other words, no one should suffer MORE because one user wishes to suffer LESS. The system is not capable of handling 100% usage by all subs at the same time. It's oversold, which is why it costs a fraction of what it would cost if it was guaranteed throughput.

You do NOT pay for full time use of your rated bandwidth 24/7. You just don't. It's not guaranteed in your TOS, nor would you have any reasonable expectation of it. Tell me you at least understand that.

I'm not insulting you, I'm attempting to get you to understand a simple concept.
--
My place : »www.schettino.us

AVonGauss
Premium
join:2007-11-01
Boynton Beach, FL
reply to fuziwuzi
said by fuziwuzi:

I am not talking about the passive capacity of "the pipe", which is irrelevant to the discussion, I am talking about active manipulation by Comcast of the effective (key word) speed.
Irrelevant? You really need to take the time to read the document explaining how the system actually works or ask questions rather than making incorrect assumptions and assertions about how the system works.


fuziwuzi
Not born yesterday
Premium
join:2005-07-01
Atlanta, GA
reply to JohnInSJ
I'm not an idiot, I am a networking professional and have been for over 20 years. YOU are the one getting caught up in semantics. You keep saying it is a simple concept yet YOU are the one trying to turn something simple into something complex.

"Packet prioritization" is just another "politically-correct" term for throttling. It is that simple. If something causes the data transfer to EFFECTIVELY be slower, then that data transfer is being throttled. That's the simple truth.

With Comcast's system, a person using the service at a time when others on their "node" are not causing the "node" be be saturated will "packet prioritize" should you use more than a certain percentage of the bandwidth they advertise for your service level. I'm sold a service that advertises 12mbps download/2mbps upload speeds, yet should I attempt to fully use that speed at ANY time, regardless of how many of my neighbors are using the service, my service will be "packet prioritized" for a period of time. That happens, I've seen it happen. If I stop using the service for a period and then resume, I will have full speeds again for a short while, then get "packet prioritized" again. However, should I limit my data transfer to around 60% of advertised speeds, I don't get "packet prioritized" and my data transfers complete as normal, though of course slower.

So you can call it "packet prioritization" if you choose, but in layman's terms it is cutting the effective speed of data transfer which translates to throttling.
--
***************
I think computer viruses should count as life. I think it says something about human nature that the only form of life we have created so far is purely destructive. We've created life in our own image.
- Stephen Hawking


fuziwuzi
Not born yesterday
Premium
join:2005-07-01
Atlanta, GA
reply to AVonGauss
said by AVonGauss:

said by fuziwuzi:

I am not talking about the passive capacity of "the pipe", which is irrelevant to the discussion, I am talking about active manipulation by Comcast of the effective (key word) speed.
Irrelevant? You really need to take the time to read the document explaining how the system actually works or ask questions rather than making incorrect assumptions and assertions about how the system works.
Please don't become condecending, I have not attacked you or made ad hominem accusations concerning your level of intelligence or understanding so I would appreciate the same professional courtesy that I have extended here.
--
***************
I think computer viruses should count as life. I think it says something about human nature that the only form of life we have created so far is purely destructive. We've created life in our own image.
- Stephen Hawking

AVonGauss
Premium
join:2007-11-01
Boynton Beach, FL
said by fuziwuzi:

Please don't become condecending, I have not attacked you or made ad hominem accusations concerning your level of intelligence or understanding so I would appreciate the same professional courtesy that I have extended here.
I'm not sure how you take my last post as condescending or as an attack on your level of intelligence or understanding. On the contrary, I've tried very hard to give you the benefit of the doubt especially when it comes to the motivation for your posts. If you have read the documents then you know how the system is supposed to work, if you feel that the system is not operating as described wouldn't it be more beneficial to post a more detailed description of what you are seeing and your environment?


fuziwuzi
Not born yesterday
Premium
join:2005-07-01
Atlanta, GA
My motivation is simply to strive for clarity. I loathe the propensity for people and corporations to "double-speak", which has sadly become the normative mode of communication in Western society.

I've explained my observations on how the Comcast system appears to operate from my perspective. I've read the documents and cutting through the "double-speak" leads me to believe that my observations are correct.

I'll make a simple analogy. You have a car that you know will go very fast. The police tell you there is no speed limit on the highway and you think, "wow, great, I can really get out there and have some fun!". However, the police also tell you that while they don't set a speed limit, if they detect you trying to exceed what they consider a safe speed, they will put water into your gasoline. Your engine will sputter and misfire and you won't be able to go as fast as you want or as fast as your car could go if otherwise unfettered.

The police say they set no speed limit, but in effect they have.

Since it is obvious we're both rather entrenched in our opinions, it is pointless to continue debating. I know what I see happening (and others see it as well). Whatever your motivation may be compels you to discredit my observations, creating an impasse.
--
***************
I think computer viruses should count as life. I think it says something about human nature that the only form of life we have created so far is purely destructive. We've created life in our own image.
- Stephen Hawking

AVonGauss
Premium
join:2007-11-01
Boynton Beach, FL
You've posted your interpretation, not the details of your actual experiences or observations. Without any of the details behind the interpretation, in my opinion, it is useless. Many of the readers here are technical by profession and/or disposition, I don't know that we really need a lot of analogies, what would seem to be more helpful and useful are the details and facts.

WernerSchutz

join:2009-08-04
Sugar Land, TX

1 edit
reply to fuziwuzi
Yes, it is double speak, although it has interesting foundations. I would call it "considerate throttling", throttling nevertheless, but an acceptable compromise. Of course, I wish it would be called what it is.


sortofageek
Runs from Clowns
Premium,Mod
join:2001-08-19
kudos:23
reply to AVonGauss
said by AVonGauss:

You've posted your interpretation, not the details of your actual experiences or observations. Without any of the details behind the interpretation, in my opinion, it is useless. Many of the readers here are technical by profession and/or disposition, I don't know that we really need a lot of analogies, what would seem to be more helpful and useful are the details and facts.
I think this is an excellent suggestion. We have read this discussion repeatedly throughout the life of this topic and, even in this subtopic, we have each heard enough to be able to make up our own minds on our choice of interpretation. I don't see any value in a back and forth of "I think it is throttling" and "No, it isn't" without a reasoned basis containing facts, particularly reasoning we haven't yet heard.

If someone feels the need to post, please let it be a post which contains the details and facts suggested instead of the same tired differences in opinion of which we grew weary long ago.
--
Join Team Helix * I am praying for these friends .


funchords
Hello
Premium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:6

1 edit
Part of the problem is that Comcast's method is a bit unique: All of the heavy user's traffic on a heavily utilized node is de-prioritized from best-effort to scavenger class for 15-minute terms.

You have to thrash the definition of throttling pretty badly to call Comcast's method, "throttling." Throttling means that you limit or manage the flow. Comcast's method does neither. If I have one complaint about Comcast's method, it's that it doesn't provide any assured floor to the affected customers. For example, I think it would be a better system than present if the flow were managed to a minimum of 128 Kbps. However, since the system so rarely kicks in (the thresholds are high and the system kicks in less than 1% of the time for any user), it's hard to complain strongly about the lack of a bandwidth floor.

QoS is connotative of using prioritization, but QoS is also connotative of guarantees of service, discrimination between different applications, meeting some specific quality goal (latency/jitter/throughput/reduced error rates). What Comcast does is more like anti-QoS because it temporarily reduces the priority of the affected customers' traffic so that they only get access to only that bandwidth that is left over by the rest of the users.

It's not throttling. It's not QoS.

"Scavenger class'' (a genuine class of service currently in use, you should take no negative connotation) allows lower-priority users and applications to take advantage of unused network capacity. It works like this: If all the normal packets have cleared the queue, then any unexpired "scavenger class" packets may be routed and forwarded. Normally, all the users crossing a particular router share the capabilities and restrictions of that connection. With a "scavenger class," the normal packets are prioritized ahead of the scavenger packets. A "scavenger class" user can use up to his full subscribed speed, if enough bandwidth is left over by other users. However, a "scavenger class" user may also find that there is no bandwidth left over by other users and that all his packets get dropped. There are legitimate uses for "scavenger class." I'm not a big fan of Comcast's use of the idea (they get around calling it "scavenger class" by using terms like "priority best effort" over "best effort" but the functional result is exactly the same). However, Comcast is both judicious in its application and has disclosed it fairly well.
--
Robb Topolski -= funchords.com =- District of Columbia -- KJ7RL
Test your Broadband connection today! -- »measurementlab.net/


JohnInSJ
Premium
join:2003-09-22
Aptos, CA
said by funchords:

QoS is connotative of using prioritization, but QoS is also connotative of guarantees of service, discrimination between different applications, meeting some specific quality goal (latency/jitter/throughput/reduced error rates).
The goal being: "always provide high availability to low-user customers." - These are the people paying the most money per bit, as it were. You want to keep those folks VERY happy. You care less about the folks that you are losing money on.

So yes, they want to maintain the Quality of Service for their cash cows.

It's QoS. You might not like how they define it, but it is.

Has anyone seen their speed drop below 128kbps during deprioritization? I've never actually experienced it myself.
--
My place : »www.schettino.us


funchords
Hello
Premium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:6
said by JohnInSJ:

said by funchords:

QoS is connotative of using prioritization, but QoS is also connotative of guarantees of service, discrimination between different applications, meeting some specific quality goal (latency/jitter/throughput/reduced error rates).
The goal being: "always provide high availability to low-user customers." - These are the people paying the most money per bit, as it were. You want to keep those folks VERY happy. You care less about the folks that you are losing money on.

So yes, they want to maintain the Quality of Service for their cash cows.
No, that's not what I mean by "specific quality goal." It's also not what I mean by "different applications."

I mean measurable and specific goals, such as, "Fairchild DBMS latency to Alpine office maintained below 70 ms. with a 95% confidence during the peak hour, and an overall average of all samples under 18 ms." This might be one of a list of goals that mention other quality vectors such as speed or packet drops. QoS Rules are the network's instructions designed to accomplish those goals.

I understand your rationale because prioritization is a tool for achieving QoS goals and we tend to think of these in connected ways, but Comcast's method is pretty far from the way we should think about QoS.

Comcast's method is a kick toward the direction of user-vs-user fairness. It could be refined more (shorter windows, better-engineered floors with more weighted queue handling) and really be an option worth considering.
--
Robb Topolski -= funchords.com =- District of Columbia -- KJ7RL
Test your Broadband connection today! -- »measurementlab.net/


espaeth
Digital Plumber
Premium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2

2 recommendations

said by funchords:

Comcast's method is a kick toward the direction of user-vs-user fairness. It could be refined more (shorter windows, better-engineered floors with more weighted queue handling) and really be an option worth considering.
It could be, but that's something even beyond the 80/20 rule in terms of effort to benefit. If there were such a thing as an 99.999/0.001 rule, this would be a quintessential case.