symmetric vs. asymmetric [was: Verizon Policy Statement on Net Neutrality]

Mark Andrews marka at isc.org
Tue Mar 3 22:57:04 UTC 2015


In message <CAMrdfRwreb_NE1zqg73V1JFXFtgRpPnNBikSd9WO8Esek13vpw at mail.gmail.com>
, Scott Helms writes:
> > I don't know many schools that are open at midnight to accept thumb
> > drives.
> 
> I think he was trying to point out that most school libraries, and their
> computer labs, open before classes start.  Ice never heard of a school
> deadline that was actually in the middle of the night, so if you're working
> on a paper at night it's because it's due the next day.

Well now you have.  See Edmodo.  The kids all have accounts.  The
teachers all have accounts.  The communication is all in the open,
no private chats.  Assignments are handed out and submitted with a
timestamp over Edmodo.

How many of you have submitted tax returns at 23:00 because you
have been running late?  Do you do this electronically now.

> > Well kids will be kids.
> >
> 
> Very true :)
> 
> >
> > Yep.  The assumption that because you are sending from home it is
> > not time critical is absolutely bogus.  Upstream speeds really are
> > just as important as downstream speeds.  It just that it is not
> > normally needed as much of the time.
> 
> This assertion is counter to the choices that consumers are making.  Forget
> about the access technology and it's symmetry or asymmetry for a moment and
> consider the growth of WiFi in the home, which is highly asymmetrical
> because clients have much lower power output and most often 0 dB gain
> antennas at 2.4 and 5.8.  The point is that a great percentage of the
> traffic we see is from asymmetric sources even on symmetrical broadband
> connections.
> The other thing to consider is that LTE is asymmetrical and for the same
> reasons as WiFi.

It is our job as engineers to give consumers what they need even if they
don't realise they needed it.
 
> For consumers to care about symmetrical upload speeds as much as you're
> saying why have they been choosing to use technologies that don't deliver
> that in WiFi and LTE?  In the WiFi case they're taking a symmetrical
> connection to their home and making it asymmetrical.  I can make a home
> WiFi network operate more symmetrically by putting in multiple APs but very
> few consumers take that step.

When you are running a nominally 400Mbps WiFi into 100Mbps fibre
you do really want to be able to fill that pipe.

> I'm not done collecting all of our data yet, but just looking at what we
> have right now (~17,000 APs) over half of the clients connected have an
> upload rate of 5mbps or less.  A just over 20% have an average upload rate
> of 1mbps.

Averages hide the peak demands.  The last mile should handle the
peak demands.  Further upstream you get the over subscription
savings.  Looking at averages and saying that they define the needs
limits is *bad* engineering.  For POTS you would get a few hertz
if you did that.  The averaging of POTS comes once you combine
multiple sources together at the exchange.  Even then you look at
the peak periods not the daily average.

Asymetry is pushing oversubscription too close to the consumer.  It
is a undesirable but sometimes necessary trade off.

Asymetry traffic volumes don't mean asymetric speeds are desirable.

> BTW, the reason we're working on the WiFi data is that we think this is a
> huge problem, because consumers don't separate the performance of the in
> home WiFi from their overall broadband experience and we need to
> dramatically improve the in home WiFi experience to increase customer
> satisfaction.

There are lots of places that need to fixed.  You address all of
them in parallel.  There are enough engineers in the world to do
that.  Just because WiFi has issues doesn't mean the next link
doesn't have issues as well.

> > Mark
> > --
> > Mark Andrews, ISC
> > 1 Seymour St., Dundas Valley, NSW 2117, Australia
> > PHONE: +61 2 9871 4742                 INTERNET: marka at isc.org
> 
> --089e010d8c2ae4b8c20510620314
> Content-Type: text/html; charset=UTF-8
> Content-Transfer-Encoding: quoted-printable
> 
> <p dir=3D"ltr"><br>
> ><br>
> > I don't know many schools that are open at midnight to accept thum=
> b<br>
> > drives.</p>
> <p dir=3D"ltr">I think he was trying to point out that most school librarie=
> s, and their computer labs, open before classes start.=C2=A0 Ice never hear=
> d of a school deadline that was actually in the middle of the night, so if =
> you're working on a paper at night it's because it's due the ne=
> xt day.</p>
> <p dir=3D"ltr">><br>
> > Well kids will be kids.<br>
> ></p>
> <p dir=3D"ltr">Very true :)<br></p>
> <p dir=3D"ltr">><br>
> > Yep.=C2=A0 The assumption that because you are sending from home it is=
> <br>
> > not time critical is absolutely bogus.=C2=A0 Upstream speeds really ar=
> e<br>
> > just as important as downstream speeds.=C2=A0 It just that it is not<b=
> r>
> > normally needed as much of the time.</p>
> <p dir=3D"ltr">This assertion is counter to the choices that consumers are =
> making.=C2=A0 Forget about the access technology and it's symmetry or a=
> symmetry for a moment and consider the growth of WiFi in the home, which is=
>  highly asymmetrical because clients have much lower power output and most =
> often 0 dB gain antennas at 2.4 and 5.8.=C2=A0 The point is that a great pe=
> rcentage of the traffic we see is from asymmetric sources even on symmetric=
> al broadband connections.<br>
> The other thing to consider is that LTE is asymmetrical and for the same re=
> asons as WiFi.=C2=A0 </p>
> <p dir=3D"ltr">For consumers to care about symmetrical upload speeds as muc=
> h as you're saying why have they been choosing to use technologies that=
>  don't deliver that in WiFi and LTE?=C2=A0 In the WiFi case they're=
>  taking a symmetrical connection to their home and making it asymmetrical.=
> =C2=A0 I can make a home WiFi network operate more symmetrically by putting=
>  in multiple APs but very few consumers take that step.</p>
> <p dir=3D"ltr">I'm not done collecting all of our data yet, but just lo=
> oking at what we have right now (~17,000 APs) over half of the clients conn=
> ected have an upload rate of 5mbps or less.=C2=A0 A just over 20% have an a=
> verage upload rate of 1mbps.</p>
> <p dir=3D"ltr">BTW, the reason we're working on the WiFi data is that w=
> e think this is a huge problem, because consumers don't separate the pe=
> rformance of the in home WiFi from their overall broadband experience and w=
> e need to dramatically improve the in home WiFi experience to increase cust=
> omer satisfaction.</p>
> <p dir=3D"ltr">><br>
> > Mark<br>
> > --<br>
> > Mark Andrews, ISC<br>
> > 1 Seymour St., Dundas Valley, NSW 2117, Australia<br>
> > PHONE: +61 2 9871 4742=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0=
>  =C2=A0 =C2=A0INTERNET: <a href=3D"mailto:marka at isc.org">marka at isc.org</a><=
> br>
> </p>
> 
> --089e010d8c2ae4b8c20510620314--
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: marka at isc.org


More information about the NANOG mailing list