W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2012

RE: fyi: An Experimental Study of Web Transport Protocols in Cellular Networks

From: <Markus.Isomaki@nokia.com>
Date: Fri, 30 Mar 2012 08:32:00 +0000
To: <ted.ietf@gmail.com>, <tobias.oberstein@tavendo.de>
CC: <ietf-http-wg@w3.org>, <g_e_montenegro@yahoo.com>, <mike@belshe.com>
Message-ID: <E44893DD4E290745BB608EB23FDDB7621B2764@008-AM1MPN1-042.mgdnok.nokia.com>
Hi,

(Sorry about hijacking HTTP-list for radio discussion, but I hope this is somehow useful background for the HTTP/2.0 work.)

Ted Hardie wrote:
>
>While the battery life issue is certainly a very important one, it's important to
>recognize that the other resource being managed here is the carrier's radio
>resources.  If you keep the device on in an active state to avoid the latency
>associated with dormant states, you are
>consuming resources that have a real impact in crowded networks. 

That's true too. I suppose it depends on the technology how much is consumed by just keeping a channel up without using it much. There is more state for sure, but the impact to other users' throughput should not be that radical. (It's a bit like having a little used RSVP reservation on a router or a TCP connection on a proxy. They consume resources but don't as such congest the data/forwarding path.) If you really wanted to cause havoc, the best way would be actually to do as many active-idle transitions as possible. That does not congest the data path either, but often the bottleneck is in those nodes who process the radio level signaling. There are cases of networs where very little goes through because of this. I think the document referenced in Ted's previous mail discusses that as well.

Markus
Received on Friday, 30 March 2012 08:32:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:57 GMT