[164535] in North American Network Operators' Group
RE: One of our own in the Guardian.
daemon@ATHENA.MIT.EDU (John van Oppen)
Mon Jul 15 04:16:37 2013
From: John van Oppen <jvanoppen@spectrumnet.us>
To: 'Jeff Kell' <jeff-kell@utc.edu>, Jima <nanog@jima.us>
Date: Mon, 15 Jul 2013 08:16:24 +0000
In-Reply-To: <51E38D85.6060702@utc.edu>
Cc: "nanog@nanog.org" <nanog@nanog.org>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
To be honest, that is the problem with most smaller ISPs, their uplinks are=
not all 10G... The only way to have users who reliably get high speed te=
sts is to make sure one does not have 1G upstream links but obviously for a=
smaller provider that would not be an option.
I think this is why our retail service routinely is in the top few on the p=
ublic speed test sites in the US... The (obvious) secret is having more t=
han 1G of headroom on every link to the world and using a lot of 10G intern=
ally. From my testing on my home link to our network and a bunch of custo=
mer links, public speed tests of above 800 mbit/sec on gigE are pretty achi=
evable assuming the testing server is in the same metro and well provisione=
d (IE not on a tiny ISP).
John
-----Original Message-----
From: Jeff Kell [mailto:jeff-kell@utc.edu]=20
Sent: Sunday, July 14, 2013 10:50 PM
To: Jima
Cc: nanog@nanog.org
Subject: Re: One of our own in the Guardian.
On 7/14/2013 9:08 PM, Jima wrote:
> XMission does offer 1000/1000, as well; I seem to recall the price is=20
> something like $300/mo. For us, the problem was more finding remote=20
> sites that can push data rates anywhere near one's own limit (as it's=20
> enough of a problem at 100mbit), making the price bump not quite worth it=
.
Very true. We have two gigs, but a commercial speedtest comes up seriously=
short (typically 100+ Mbps) while a locally hosted speedtest will show 800=
-900+. Not sure how much is "their" upstream versus simple physics... you'=
d have to be the only test subject to a gig-connected server to do much bet=
ter.
We have had some "contrived" examples over I2 that pushed 500Mbps symmetric=
, but they ran that demo over our I2 pipe because their commodity link coul=
dn't deliver the necessary rate/latency.
Jeff
