>>> I think perhaps something is just a bit broken in the new DSLAM firmware because it still sort of works. I suspect that the "upgrade" may have had more to do with enforcing the 16Mbps "limit" that the service has now.
>> There has to be a significant financial reason for that I think as the savings in support costs are obvious but........
I cant quite figure that out ... aside from continually running at max could obviously cause some problems - but surely there are still other configurable aspects such as target SNR?
"In theory 33:1" should be ok - but obviously it depends on the type of user at your particular exchange. Unfortunately those who elected for the higher speeds are probably most likely to be higher users
>> I should start monitoring latency a bit more
You know your BTw exchange is one of the very few which Ive seen suffer from latency problems due to contention (ignoring all the long grass stuff).
In all the times that Ive suffered contention despite from some very low speeds Ive never had a problem with latency nor did the other 30+ users on our exchange. Ditto with the 1Mb and 2Mb upgrades. Poor speeds but latency ok. About 12-18 months ago when about 60% plus exchanges were on red I was monitoring AG and although poor speeds were an obvious issue I never saw anyone complaining about latency.
I suspect that some dslams are capable of doing a basic type of QoS and BT do use it for stuff like latency. IIRC you were on one of the early enabled exchanges with possibly an older type dslam. The Juniper ERX's installed by BTw during the period circa 2003-2005 certainly is capable of doing a basic form of traffic prioritisation. Its were the gold/silver/bronze stuff came from in the first place.
I was told in 2004 that its the older CISCO dslams that have problems identifying traffic/account types and hence why BT replaced a load of CISCO dslams in 2004 with Junipers.
>> *By "links" I mean rented transit rather than "We own the fibre"
Still cheaper than centrals though