[GRLUG] AT&T's U-verse DSL service

[Bob Kline]

No extreme performance issues beyond
apparently daily load variations in response
and speed.

Re no dedicated connection, by that I
mean that I don't to the same extent
share a cable with other users at the
local level.  i.e., I've heard two different
figures for how Comcast allocates capacity.
One says 40 Mbps per 300 users, and another
says 50 Mbps for 400 users.  With DSL,
one supposedly gets to the CO over one's
own connection, like the old phone approach.
What happens in the CO is the question.
Perhaps there's just congestion at that
point.  That's really the question - whether
there is the same concept of loading with
the DSL.  If many people use their cable
modems on the same cable, the capacity
can be used up.  Same true with DSL?

This is different from server congestion,
and I presume there is little backbone
congestion.

Anyway, some of this can be answered
by using something like speedtest.net.
I see big variations throughout the day.
My questions is whether people with
high speed DSL connections see similar
variability.  I can upload and download
big files to my web hosting site, and
simply measure the time.  That's a direct
measure of rate.  Same for downloading,
say, a new distribution of Ubuntu.

At bottom, small variation, for any mean
speed, is good - for me.  That way I know
whether I need more, and what to plan on.

    -- Bob


On Tue, May 17, 2011 at 10:18 PM, Adam Tauno Williams <
awilliam at whitemice.org> wrote:

> On Tue, 2011-05-17 at 21:06 -0400, Bob Kline wrote:
> > A few people in the group have AT&T's 24 Mbps
> > U-verse DSL service.  I'm interested in knowing
> > how the service behaves.
>
> I'm a very happy U-verse customer; with a much lower plan than 24Mbps.
>
> > e.g., Comcast's services clearly varies with load
> > throughout the day, and one sees factors of as
> > much as 4 in the bite rate in tests on speedtest.net
> > ( Just for reference, Comcast sponsors the Chicago
> > test point, and always shows the highest rates... )
>
> Speaking as someone who has access to an OptE-man Fiber connection...
> these speed tests are meaningless and pointless.
>
> Do you have performance issues?  That is the only relevant question.
>
> There is mean throughput, peak throughput, and latency.  Any meaningful
> test will benchmark all three, will need to run for hours, and you'll
> need to simulate a real-world mix of network traffic.
>
> > So, does AT&T's service act more like a dedicated
> > connection,
>
> There is no such thing as a 'dedicated connection' on the Internet.
> Nothing will act like one [from someone who has many point-to-point
> circuits as well as a private MPLS cloud].
>
> I see no evidence of last-mile contention on my U-verse circuit.
>
> > like one more less fantasizes a DSL
> > connection could have, since you have a "wire"
> > right back to the CO?  i.e., does anyone know
> > whether that is the case?  Many things come in
> > to play here, including congestion on the bigger
> > "wires," but I'm pretty much assuming that test
> > server congestion is not an issue.
>
> A big assumption, even after one assumes their test means anything.
>
> > Of course service congestion is a time of day
>
> Maybe, maybe not.  I suspect usage patterns are actually quite complex
> and influenced by everything from time of day, to weather, to current
> events.
>
>
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
>
> _______________________________________________
> grlug mailing list
> grlug at grlug.org
> http://shinobu.grlug.org/cgi-bin/mailman/listinfo/grlug
>

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://shinobu.grlug.org/pipermail/grlug/attachments/20110517/2db7e5d5/attachment-0001.html>