[142520] in North American Network Operators' Group
Re: [pfSense Support] Strange TCP connection behavior 2.0 RC2 (+3)
daemon@ATHENA.MIT.EDU (Cameron Byrne)
Wed Jun 29 16:53:49 2011
In-Reply-To: <alpine.DEB.2.00.1106292247410.19581@uplift.swm.pp.se>
Date: Wed, 29 Jun 2011 13:53:44 -0700
From: Cameron Byrne <cb.list6@gmail.com>
To: Mikael Abrahamsson <swmike@swm.pp.se>
Cc: NANOG list <nanog@nanog.org>,
"williamejsalt@googlemail.com" <williamejsalt@googlemail.com>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
On Wed, Jun 29, 2011 at 1:48 PM, Mikael Abrahamsson <swmike@swm.pp.se> wrot=
e:
> On Wed, 29 Jun 2011, Cameron Byrne wrote:
>
>> Sounds like a vendor specific issue :(
>
> Absolutely, but this is way too typical for these kinds of networks.
>
>> Good tcp vs default tcp will not close the window tight due to some
>> ephemeral loss or delay. =A0The penalties are generally too strong in tc=
p
>> for
>> the issues of delay and loss in 3g ... this is one of the main selling
>> points for tcp proxies .... but better done with modern tcp on the clien=
ts
>> instead of a middle box
>
> For what kind of devices? I can get full speed (870 kilobyte/s) on 7.2 HS=
PA
> with single TCP stream on a linux box. Are you referring to handsets that
> get improvement on high latency links?
>
Yes. Smartphones on UMTS/HSPA have poor bandwidth-delay product,
hence it being common for mobile providers to proxy the TCP to open
windows faster.
Cameron
> --
> Mikael Abrahamsson =A0 =A0email: swmike@swm.pp.se
>
