Re: [gentoo-user] Re: Wifi slow motion data transfer

Mick <michaelkintzios@×××××.com> [15-07-29 20:16]:
> On Wednesday 29 Jul 2015 18:38:07 James wrote:
> >  <Meino.Cramer <at> gmx.de> writes:
> > > J. Roeleveld <joost <at> antarean.org> [15-07-29 16:38]:
> > > > On Wednesday, July 29, 2015 05:18:25 AM Meino.Cramer <at> gmx.de wrote:
> > > >
> > > > Is this speed between both machines? Or to the internet?
> >
> > Joost is exactly correct here. Test the links one connection at
> > a time, not just all a once. You'll be able to get a 'sit of pants' feeling
> > about the capabilities on each link (between devices). There are many many
> > issues so let's first characterize each link by the bandwith.
> >
> >
> > On ethernet interfaces this is a really cool tool::
> >
> >  net-analyzer/bwmon   and net-analyzer/nbwmon
> >
> > > I fired up create_ip like this (just for testing and haveing at least
> > > ONE experienced succes with this Wifi stuff...):
> > >
> > > create_ap wlan0 eth1 <name> <pass>
> > >
> > > How can I check for the type of WIFI after the connection has been
> > > established?
> >
> > 'ip link'   and  'netstat -nr' are a good start. Later on we'll get
> > you some gui tools and a monitoring software (a ton of options)...
> >
> > > USB is USB 2.0
> > >
> > > The speed is measured by conky, which reads the transfer rate at eth0.
> > > At that tome, the tablet was getting a greater piece of tar archive
> > > (LInux for Android) and no other traffic other than this was there.
> > > The DSL was by far not saturated.
> >
> > Really?  How do you know. It take lots of experimenting and testing
> > and data collection over time to figure our exactly what your
> > ISp(s) are doing. Usually several ISPs are in a link until you hit
> > a 'peering point'
> >
> > 'net-analyzer/traceroute'
> >
> > is your friend. At some point the ISPs will block traceroute info....
> >
> > > So physically it is the speed of the internet but logically it is
> > > nearly identical to what happens at the Wifi interface (I think).
> > > I will check for an app which displays the speed measured on the
> > > tablets interface...
> >
> > This is a very, very complicated issue. ISP(s) use devices to deliver
> > and partition bandwidth; some with an incredible level of control
> > (granularity). For instances they can 'port constrict' a service
> > or a route to an endpoint or any number of things. So first fully
> > study (characterize) the behavior of the links (connnections between
> > devices) that you manage and develop that 'seat of pants' feeling about the
> > network segments you manage. Then start sniffing up the outside folks,
> > as best you can with the tools in the portage tree.....(many).
> >
> >
> > You need to also understand that Usb has it's own problems, protocols and
> > issues depending on how it was implemented by the chipsets use and the
> > firmware inside the product. Other protocol (latencies and such) are
> > layered on top of that.  Ju are 'full stack' wheelin and dealing as soon
> > as your run gui apps across that link.........brau.
> >
> > > Best regards,
> > > Meino
> >
> > ttfn,
> > Always your pal!
> > James
>
> It could also be that the tablet has a slow write speed, if you were
> downloading a file.  Can you stream a video instead and see if this is
> achieving a higher speed?
>
> --
> Regards,
> Mick


Hi Mick,

thanks for your infos! :)

yupp! Works like a charm (video streaming)!

It was a default setting in create_ap which seems to select an old
procotol based on drums and morse code... ;)
After adding --ieee80211n to the commandline the problem vanished
and the data transfer rate jumps up to a level, where the limiting
device was my DSL provider :)

I am happy with that now !

Best regards,
Meino