Rogers explains ‘shaping' policy

James Knott james.knott-bJEeYj9oJeDQT0dZR+AlfA at public.gmane.org
Thu Jun 12 22:11:20 UTC 2008


CLIFFORD ILKAY wrote:
> You sound like a Roger PR guy.
>
> Stephen wrote:
>> D. Hugh Redelmeier wrote:
>>> The question is how they should divy up the bandwidth.
>>>   
>> I thought that this was addressed well.
>>
>> They divvy up the bandwidth by partitioning the residential bandwidth 
>> so most is for downloading,
>> and little is for uploading.
>
> What is so special about uploaded vs. downloaded packets? Are uploaded 
> packets "heavier"? Bandwidth is bandwidth. I should be able to specify 
> the QoS so that I can dynamically allocate the proportion of up/down 
> rates as my needs change. All Rogers has to tell me is that I can 
> burst up to some transfer rate. I'll take care of the rest.

1) Most users download far more than they upload, ISPs adjust their 
network accordingly.
2) Upload bandwidth is in a different part of the spectrum, below the TV 
channels, where there is less bandwidth available.  Download has much 
more space available.  This is determined by filters on the cable plant, 
which separate the directions.
3) Contention.  There has to be some mechanism to ensure fair access for 
everyone.  This results in some inefficiency of the bandwidth use.
On the other hand, download data is just sent out as quickly as possible.

Points 2 & 3 are caused by the coaxial cable plant, used for the "last 
mile".  They do not apply to the fibre optic cable used for the trunks.  
Point 2 also applies to ADSL connections.  While it is possible to get 
symmetrical service, it costs more than ADSL and the bandwidth is less 
than the download ADSL bandwidth.  Again this is due to cable limitations.



For more info on this, google on DOCSIS specs.

-- 
Use OpenOffice.org <http://www.openoffice.org>
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists





More information about the Legacy mailing list