"Phil Thompson" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> On 17 Dec 2004 06:50:40 -0800, (E-Mail Removed) (hals_left) wrote:
>
> >and broadband all seems to be high contentions.
>
> you have to ask what contention effect is seen in practice. ADSLguide
> speed tests, for example, show averages of fairly large samples up at
> around 90% of rated speed.
Agreed. and since uplinks run slower but the DSLAM backhaul is symmetric
bandwidth (and uplinks tend to be less heavily loaded) - the effective
contention on uplink is better.
>
> Rumour is that 50:1 Home products are actually contended at more like
> 15:1 in practice.
this may well be due to the large number of DSLAMs. Each ISP buys job lots
of backhaul per DSLAM - so if they only have 7 users on that DSLAM, 7:1 is
as poor as the ISP can make it.....
Evan then the "average" line spends most of its time idle, so you get a lot
of statistical gain.
A pair of 2M 20:1 ADSL circuits into a load
> balancing router would probably be fine for 25 users assuming they
> aren't all maxing out their connection continually, having 2 circuits
> gets you 512k of upload to share, but one user would only see 256k max
> (unless you went for 2 bonded circuits).
This is true so long as the ISP has enough users that you dont have your 2
circuits contending with themselves all the time.
If you 2 * 2M feeds, and the DSLAM is configured with a single 2M aggregate
for your ISP and that ADSL speed, then you wont get much extra throughput
over a single ADSL link.
>
> Phil
> --
> spamcop.net address commissioned 18/06/04
> Come on down !
--
Regards
Stephen Hope - return address needs fewer xxs