13 1998 (ftp://thumper.bellcore.com/pub/huitema/stats/quality_tod

Stefan Savage (savage@cs.washington.edu)
Tue, 20 Oct 1998 15:51:35 -0700

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------ =_NextPart_000_01BDFC7C.33E4B8F6
Content-Type: text/plain

This is similar to the measurements that a number of us are interested
in seeing. Of course, there's no information about methodology,
details, etc... its especially unclear how server delay is separated
from transmission delay.

However, if you buy these numbers, then a reasonable amount of the time
we're being delayed by servers and not queuing. On the other hand, look
at the median packet loss rate of 16%! Ouch. Its also nice to see that
DNS delay is substantial (I've always suspected that DNS is a problem
child). Finally, it is totally unbelievable to see the difference
between the median page delay and the sum of DNS, connection, server and
transmission delay (more than double). I can't believe this is all
rendering and layout time... if the page has multiple inlines then it
could be that this reflects exponential backoff on one of the several
connections for the page (the variance on the connection and
transmission delay is quite high and the sum of the averages works out).

- Stefan

<<13 1998.htm>>

------ =_NextPart_000_01BDFC7C.33E4B8F6
Content-Type: application/octet-stream;
name="13 1998.htm"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: attachment;
filename="13 1998.htm"

<BASE =
HREF=3D"ftp://thumper.bellcore.com/pub/huitema/stats/quality_today.html"=
>=0A=
=0A= =0A= =0A= Internet quality of service assessment,=0A= Tue Oct 20 15:33:13 1998=0A= =0A= =0A= =0A=

=0A= Internet quality of service assessment,=0A= Tue Oct 20 15:33:13 1998=0A=

=0A= In order to assess the quality of our Internet service, we have drawn = at =0A= random a set of =0A= 118 URL (that is, addresses of web servers.) =0A= 18 connection attempts have failed.=0A= For the remaining =0A= 100 connections, we have measured the delay that we had to wait for = loading=0A= the page in four components:=0A= =0A= The following table provides the median, 95th percentile and average = value=0A= for these components, as well as for the complete page delay.

=0A= =0A= =0A= =0A= =0A= =0A= =0A= =0A= =0A= =0A= =0A= =0A=
Median95 perc.Average
DNS delay (ms)32340881377
Connection delay (ms)1506301797
Server delay (ms)261180664211
Transmission delay (ms)708186995520
Page delay (ms)37354001811907
=0A=

=0A= When we took the measurement, we also carefully analyzed the behavior = of the=0A= TCP connection. This gives us an indication of the round trip time, of = the=0A= available data rate and of the packet loss rate. because our pages = have been=0A= selected at random, these figures are an indication of the state of = the=0A= Internet. The values that we found today are detailed in the following = table:=0A=

=0A= =0A= =0A= =0A= =0A= =0A= =0A= =0A=
Median95 perc.Average
Round trip time (ms)1181151328
Transmission speed (bps)46376161699678
Packet loss rate (%)163315
=0A=

=0A= =0A= =0A=

------ =_NextPart_000_01BDFC7C.33E4B8F6--