Comment 21 for bug 582157

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Martin Pool wrote:
> @John the most obvious difference between your machine and Vincent's is
> the different round trip time, which connects to the TCP windowing
> restriction discussed above. What ping time do you see to chinstrap
> from those two machines?

It seems to depend on time of day. Vincent tested after seeing my
original mail, and said he, indeed, saw 20ms. We just tested again right
now (17:00 UTC) and we both see 120ms to chinstrap or bazaar.lp.net.

Actually, at this instant, my local connection is at 110ms (so actually
faster than Vincent's.)

And while it does do a little bit better job saturating the bandwidth, 2
concurrent downloads is flatlined at cap, 1 is still 'bouncy'. (I made
sure to test from my server which has a wired connection, to eliminate
any noise from my house wireless.)

So even though the ping times are comparable, I certainly see far more
download speed to Vincent than myself. (He has a bigger pipe, but you
would think we could fully saturate my pipe with a single download. I
*do* see a full saturation downloading a single tar.bz2 from kernel.org.)

IOW, it doesn't seem to be strictly linear vs ping time.

Then again, my ping time never gets as low as 20ms, so it may be
something at this time of day that de-prioritizes ICMP?

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkwqMEoACgkQJdeBCYSNAAMkigCgjXWVNfxPxH9IKAIU+FZ8SuqQ
G3cAn2A6J22uupuicLoopCx74xPajBgp
=ytVz
-----END PGP SIGNATURE-----