[Codel] CoDel and GPRS in osmo-pcu

[Codel] CoDel and GPRS in osmo-pcu

[Jacob Erlbeck]

Hello,

just for your information:

we have added active queue management based on CoDel to the osmo-pcu
project (https://openbsc.osmocom.org/trac/wiki/osmo-pcu), which is an
open source software based GPRS-only PCU implementation.

The implementation is based on
http://queue.acm.org/appendices/codel.html but is deployed as a time
stamp based automaton. It is being applied independently per mobile
station (phone) in downlink direction only.

The goal was to have PCU based AQM as a last resort if BSSGP flow
control is not sufficient. Since the IP packets are eventually split and
encapsulated within LLC frames which are generally encrypted, the PCU is
not an optimal place to do AQM. Nevertheless it improves the TCP
"experience" and downlink ping latencies under load significantly.

Kind regards

Jacob

-- 
- Jacob Erlbeck <jerlbeck@sysmocom.de>       http://www.sysmocom.de/
=======================================================================
* sysmocom - systems for mobile communications GmbH
* Alt-Moabit 93
* 10559 Berlin, Germany
* Sitz / Registered office: Berlin, HRB 134158 B
* Geschaeftsfuehrer / Managing Directors: Holger Freyther, Harald Welte

Re: [Codel] CoDel and GPRS in osmo-pcu

[Kathleen Nichols]

Cool.

On 12/21/15 12:58 AM, Jacob Erlbeck wrote:
> Hello,
> 
> just for your information:
> 
> we have added active queue management based on CoDel to the osmo-pcu
> project (https://openbsc.osmocom.org/trac/wiki/osmo-pcu), which is an
> open source software based GPRS-only PCU implementation.
> 
> The implementation is based on
> http://queue.acm.org/appendices/codel.html but is deployed as a time
> stamp based automaton. It is being applied independently per mobile
> station (phone) in downlink direction only.
> 
> The goal was to have PCU based AQM as a last resort if BSSGP flow
> control is not sufficient. Since the IP packets are eventually split and
> encapsulated within LLC frames which are generally encrypted, the PCU is
> not an optimal place to do AQM. Nevertheless it improves the TCP
> "experience" and downlink ping latencies under load significantly.
> 
> Kind regards
> 
> Jacob
>