ADSL caps punishing the wrong subscribers?

Interesting read... I wonder if this is what webafrica are trying to address with their 6 hour and 7 day thresholds?
 
Basically what they saying is that Very Heavy users aka Hogs are limited during peak times and therefore can not be blamed for poor network performance. Yet poor network performance exists during peak times and this is attributable to the hisher speed lines.
 
My head hurts after reading all that, would love to see some graphs to illustrate their point .. but then I suppose that'll be what's in the paid version. :cry:

It does sound like they say caps were designed to protect the network against the couple of highest intensity hours of the day, which is understandable. But, that there's scope for increasing bandwidth allocation as long as the extra usage can be spread accross the lower peak period.

Amirite?
 
Basically what they are saying is this:

- There are two things you can measure: Data Usage (how much you download) and Bandwidth Usage (how fast you download)
- The ISPs are worried about Bandwidth Usage
- Capping Data Usage is not going to affect Bandwidth Usage
- ISPs don't actually know how their networks are being used
 
Last edited:
Is this not exactly what SA ISPs like Mweb are doing by shaping during the day? Meaning you can download as much as you want as long as you don't do it during peak times.
 
TheGuy said:
Is this not exactly what SA ISPs like Mweb are doing by shaping during the day? Meaning you can download as much as you want as long as you don't do it during peak times.
Click to expand...

Pretty much which is why uncapped with various shaping and throttling options (i.e. cheaper to more expensive) is the way to go. Capping is pretty pointless if everyone who is capped decides to access the internet at the same time as the bandwidth becomes limited and everyones' experience degrades.

What was kind of worrying was the mention that some ISPs did not actually measure properly so could not give data:wtf: How can you run an ISP and not monitor data throughput:confused:
 
Bern said:
Pretty much which is why uncapped with various shaping and throttling options
Click to expand...
But I think the 2 options are seen differently. My take is these guys are saying using consumption (GBs used) is an ineffective way of managing peak aggregate bw demand.

Since throttling uses consumption (thresholds) while shaping does not (occurs in realtime, all the time), I suppose you could infer they believe a mechanism like shaping would be more effective than throttling.
 
"Finally, one such ISP, a mid-size company from North America, agreed to share a data set for analysis."

"the highest service tier (6 Mbps)" "3Mbps (the next service tier)"

http://en.wikipedia.org/wiki/Internet_in_the_United_States

"the United States ranked 26th globally in terms of the speed of its broadband Internet connections with an average measured speed of 4.93 Mbps"

So, with an average of a 5MB Line, this must be a pretty low-end company to have their "Highest" tier being a 6MB Line? (I also find it odd how Telkom apparently has faster internet than this specific US company... ? :confused::confused::confused:)

"Thisroughly equates to data consumption of 8.7 GB and 288 GB per month, respectively."

I've seen many MyBB users report 1TB+ DL totals / month - I find it odd that the "High Tier" users on these 6MB Lines only manage <300GB....

"That report is for sale, but as promised I reproduce the executive summary here."
"users’"?, "worldview", "setwas", "Thisroughly"

For a costly report, I find the quantity of typo's in this small paragraph disturbing :p
 
Last edited:
Bern said:
What was kind of worrying was the mention that some ISPs did not actually measure properly so could not give data:wtf: How can you run an ISP and not monitor data throughput:confused:
Click to expand...

Actually I think it is not to do with bandwidth throughput graphs. Anyone can set those up, and trust me, they don't tell an ISP what they want to know. There are some better tools, like netflow that can give you a bit of a better idea, but I think the best way to really know what your users do, is DPI (deep packet inspection), look at the actual protocols and where these protocols are going. Once you know that, you start setting up peering links, maybe even peer-to-peer caching devices, if you traffic is mainly HTTP and not peer-to-peer, hell then a peer-to-peer caching device become useless real fast. What if your main traffic on your network is streaming? That is mostly not cache-able unless you have something like Google Cache servers.

Nowadays, something I would like to test, is a device that recognizes patterns (ie. like in video streaming) and then instead of downloading the patterns the whole time, send it from a server which already has seen these patterns, ie. almost like caching a video stream, except this happens in real time.
 
Reelix said:
"Finally, one such ISP, a mid-size company from North America, agreed to share a data set for analysis."

"the highest service tier (6 Mbps)" "3Mbps (the next service tier)"

http://en.wikipedia.org/wiki/Internet_in_the_United_States

"the United States ranked 26th globally in terms of the speed of its broadband Internet connections with an average measured speed of 4.93 Mbps"

So, with an average of a 5MB Line, this must be a pretty low-end company to have their "Highest" tier being a 6MB Line? (I also find it odd how Telkom apparently has faster internet than this specific US company... ? :confused::confused::confused:)

"Thisroughly equates to data consumption of 8.7 GB and 288 GB per month, respectively."

I've seen many MyBB users report 1TB+ DL totals / month - I find it odd that the "High Tier" users on these 6MB Lines only manage <300GB....

"That report is for sale, but as promised I reproduce the executive summary here."
"users’"?, "worldview", "setwas", "Thisroughly"

For a costly report, I find the quantity of typo's in this small paragraph disturbing :p
Click to expand...

They are probably an ISP not in, but close to a major centre. The US has the same problem here that the less condenced areas either have no access or very low speed internet access. Averages like this are never a good indication of whats available.

I would like to know what the data was that everybody is downloading. For example theirs (USA) could consist a lot of Hulu and netflix traffic and ours (SA) might be totally different. And I've only heard of some mybroadband users thats doing terabytes and I dont think they are a representation of what all users on the network do.
 
The authors assume the ISP's input costs are directly related to bandwidth capacity, and totally unrelated to consumption. This isn't necessarily the case, and in fact isn't the case in SA for any 'SAIX ADSL resellers' (aka any ISP which doesn't use IPC). As such, their conclusion isn't valid for all ISPs.

In conclusion, we state that policies honestly implemented to reduce bandwidth usage during peak hours should be based on better understanding of real usage patterns and should only consider customers’ behavior during these hours; their behavior when the link isn’t loaded cannot possibly impact other users’ experience or increase aggregation costs.
Click to expand...
.

If the ISP is billed directly by their wholesale provider based on the total consumption of their users, then capping implements a direct revenue-cost control.
 
Very interesting piece - and I'm glad they addressed the difference between MB and Mbps. However, I'm left with one question.

It is apparent from their findings that capping the consumption, when they should just be throttling the rate of consumption. What worries me is that data caps force a certain behaviour. When there are no caps, how would this alter the behaviour of heavy data users?

Take Mweb's R399 deal a while back. Everyone went nuts and dl'd like crazy, effectively rendering the service unusable. Having said that, I do think it would be the same as legalising drugs. The first few months people would go off their rockers at this new found proverbial pot of gold, but a few months in they will have grown accustomed to the availability of bandwidth/data (our drug in this case). We will no longer be maxing our lines with downloads, but rather use it "on demand".

The real trump card will be the effect of popularising live streaming which is both bandwidth and data intensive. Who is going to be the first ISP to commercialise set top boxes on a national scale?
 
What is "setwas", "Thisroughly", "isabove" and "onepoint"? Is it broadband terms?
 
Last edited:
r00igev@@r said:
They critise ISPs and they themselves don't look at latency, connection count and packet size. They are as one-dimesional as they ISPs themselves.
Click to expand...

Perhaps they are taking baby steps, first get some basic information and then build on that. Remember ISPs probably don't want to share too much in case they look bad.
 
Voicy said:
The real trump card will be the effect of popularising live streaming which is both bandwidth and data intensive.
Click to expand...

IMHO, the obstacle to live streaming at the moment is a different monopoly (in terms of licensing content) and their anti-competitive behaviour, not infrastructure ...
 
The_Librarian said:
Also "I'm too lazy to check the final product for mistakes" :whistle:
Click to expand...

No, he probably just copied and pasted the executive summary from the paper into his blog (since that's pretty much what he said he did), and those were the result of the copy-paste. It's happened to me before on numerous occasions.

That report is for sale, but as promised I reproduce the executive summary here.
Click to expand...

Also just realise that English is not his first language, he's actually French.
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X