On 7 Jun 2012, at 19:49, Neil J. McRae wrote: > in some cases yes - ATM based DSL lines are particularly bad at coping with > VoIP under congestion - I presented some stats from my days at COLT on this > at UKNOF4 I think. >
buffering? > However even with QoS and other priority chaos there is still no guarantee > even on IP friendly access links. > Well QoS is the mother node for misaligning network capacities with user applications! So I guess this is an understatement. But to be clearer I was taking your statement at face value that capacity in a network allows best effort IP services without resorting to QoS. That seems to be mainstream IP architecture thinking. I see users (called broadband consumers by some) who are using apps like video chat with HD pretty much on all day to multiple sites in a similar way IRC was used for early adopters and this doesn't seem to be in the "game plan" assumptions behind provisioning Async tails to them as the default when performance is limited to the slower upstream rate. As things seize up at the edge for these users they have to reset and or the ISPs throttle (or QoS sic!) to cope but these 'solutions' are hardly market, application or network friendly. This and other changes in behaviour make me wonder how long async paths at the user edge is the sensible default plan. If the "architecture" is to have sufficient bandwidth to avoid need for QoS then should that not be bidirectional and include the user edge? best Christian > Sent from my iPad > > On 7 Jun 2012, at 16:10, "Christian de Larrinaga" <[email protected]> wrote: > >> Neil >> >> Do you find the model of asynchronous links to premises makes it tougher for >> everything to get "enough bandwidth" ? >> >> Christian >> >> >> >> On 7 Jun 2012, at 15:05, Neil J. McRae wrote: >> >>> IP networks 101. >>> >>> If you have enough bandwidth everything gets a good deal. >>> >>> Sent from my iPhone >>> >>> On 7 Jun 2012, at 10:42, "Justin Finkelstein" <[email protected]> wrote: >>> >>>> nteresting; so it sounds, from what you say, that by default VoIP gets a >>>> "good deal" out of the available bandwidth? >>>> >>>> On 7 June 2012 09:55, Adrian Kennard <[email protected]> wrote: >>>> On 07/06/12 09:45, Justin Finkelstein wrote: >>>> > >>>> > This may be the wrong forum for this message, but I was wondering about >>>> > how VoIP is handled as part of people's various traffic shaping policies >>>> > - i.e. is this something people take account of? >>>> >>>> Our LNSs know the rate of every line, and shape to match the line rate >>>> but with small packets having more priority over large - this works well >>>> for VoIP, but also ACK packets, DNS, interactive (key strokes), etc. We >>>> find people can fill their line with torrents and still have perfect >>>> VoIP calls. >>>> >>>> > Also: if one of the UK's ISP's decided to implement QoS on their >>>> > customer base's users and then prioritise VoIP, what kind of impact >>>> > would this have on the upstream network? >>>> >>>> Upstream is not usually an issue as it is not usually congested! >>>> >>> >> >
