[149061] in North American Network Operators' Group
Re: 10GE TOR port buffers (was Re: 10G switch recommendaton)
daemon@ATHENA.MIT.EDU (bas)
Fri Jan 27 17:54:25 2012
In-Reply-To: <596B74B410EE6B4CA8A30C3AF1A155EA09C965B5@RWC-MBX1.corp.seven.com>
Date: Fri, 27 Jan 2012 23:53:35 +0100
From: bas <kilobit@gmail.com>
To: George Bonser <gbonser@seven.com>
Cc: "nanog@nanog.org" <nanog@nanog.org>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
While I agree _again_!!!!!
It does not explain why TOR boxes have little buffers and chassis box
have many.....
On Fri, Jan 27, 2012 at 11:36 PM, George Bonser <gbonser@seven.com> wrote:
>>
>> Buffers in most network gear is bad, don't do it.
>>
>
> +1
>
> I'm amazed at how many will spend money on switches with more buffering b=
ut won't take steps to ease the congestion. =A0Part of the reason is trying=
to convince non-technical people that packet loss in and of itself doesn't=
have to be a bad thing, that it allows applications to adapt to network co=
nditions. =A0They can use tools to see packet loss, that gives them somethi=
ng to complain about. =A0They don't know how to interpret jitter or underst=
and what impact that has on their applications. =A0They just know that they=
can run some placket blaster and see a packet dropped and want that to go =
away, so we end up in "every packet is precious" mode.
>
> They would rather have a download that starts and stops and starts and st=
ops rather than have one that progresses smoothly from start to finish and =
trying to explain to them that performance is "bursty" because nobody wants=
to allow a packet to be dropped sails right over their heads.
>
> They'll accept crappy performance with no packet loss before they will ac=
cept better overall performance with an occasional packet lost.
>
> If an applications is truly intolerant of packet loss, then you need to a=
ddress the congestion, not get bigger buffers.
>
>
