On Tue, Mar 9, 2010 at 12:04 PM, Quentin Wright <[email protected]> wrote:
> Darren Worrall wrote:
>> Hi all,
>>
>> Some notes have been posted about the network at this years Pycon:
>> http://www.tummy.com/Community/Articles/pycon2010-network/
>>
>> I'm not sure how much of it is relevant to us, or how it compared to
>> last years EP network, but it's worth a read by whoever has
>> responsibility for the network this year - the wireless breakdown is
>> pretty interesting. Headlines: Half of the audience were using
>> 802.11n, 5.2GHz scales better and is much more reliable than 2.4GHz.
>>
> I've been responsible for the tech team at PyConUK and EuroPython for
> the past
> few years.
>
> Initially two bonded ADSL lines were installed by a Guadec Conference
> and taken
> over by us and subsequently the Conservatoire.
>
> At the beginning all cabling and positioning of access points was ad
> hoc, and
> in-fill where coverage was lacking was made during the event.
>
> In recent years the Conservatoire has taken over the infrastructure and
> placed
> access points within ceilings etc.
>
> Last year we were not allowed to have access to the comms room, were not
> given
> passwords to the routers and were not allowed to put flying ethernet
> cables in
> place, and could not fill in holes and in particular couldn't put extra
> access points in the Adrian Boult Hall. All this lead to much frustration.
>
> For this year we are planning to have a completely new set-up with our
> own broadband lines, 802.11n access points and possibly a mesh network,
> or failing that wireless bridging. Ideally this will be independent of
> the Conservatoire's infrastructure.
>
> The PyCon 2010 comments on the reliability of 2.4 GHz are not really
> accurate because as far as I am aware they are still configuring their
> access points with unique essid's, rather than having a single essid
> common to all access points. This means that roaming doesn't work
> properly and in particular the number of dchp requests is excessive.
This is incorrect. Unique ssid's were only set up in reaction to
problems, and were set up for 802.11a on the 5Ghz frequency.
All of the 2.4 apps were using the same 'pycon' ID.
We peaked at around 500 unique people on the wireless, and still
saturating the apps (64 per app). It turns out this is due to those
pesky iPhones which will attempt to connect to all the apps with the
same ID, and use up a slot on all the nearest radios. They would
appear to swap out and not stay connected, but this was causing major
problems. Thus the creation of special extra ID's for people to use.
-Doug
> More people in the USA have new machines with 802.11n because the price
> of hardware in real terms is much less than in the UK.
>
> Additional thoughts and assistance is welcomed!!
>
> Cheers,
>
> Q
>
>
>
> _______________________________________________
> Europython-improve mailing list
> [email protected]
> http://mail.python.org/mailman/listinfo/europython-improve
>
_______________________________________________
Europython-improve mailing list
[email protected]
http://mail.python.org/mailman/listinfo/europython-improve
