Re: Adtran async/sync DSU 5600 with PM's
John Capo (jc@irbs.com)
Thu, 5 Oct 1995 00:59:59 -0400 (EDT)
Michael Dillon writes:
>
> On Thu, 28 Sep 1995, Dick St.Peters wrote:
>
> > I wrote:
> > > Steve Davies writes:
> > > > You are misinformed about V.34/V.34+ on two-wire. They are full-duplex.
> > >
> > > They are half-duplex.
> >
> > I have to aknowledge being wrong.
> >
> > That leaves the poor peformance of a 28.8 dialup compared to half that
> > of a 56k line unexplained. Modem turnaround time is the explanation I
> > have often seen given.
>
> My theory is that it is like a highway. On a four-lane highway (i.e. two
> lanes in each direction) you get a lot more than twice the traffic
> throughput than a two-lane highway. I'm sure there is some mathematics of
> queuing theory that explains why this is, but seeing as how an IP channel
> is actually a whole bunch of multiplexed conversations this could explain it.
>
> The other factor is that when you put a bit into a 56K sync line it
> immediately appears at the other end but when you put a bit into a modem
> there are buffers and v.42bis compression doing stuff which slows it down
> somewhat.
>
I just switched from v.42bis compression in the modem to predictor-1
compression outside the modem and shaved 30Ms off the ping time of
a 21600Bps DCE link.
John Capo
IRBS Engineering High performance FreeBSD systems