[Cerowrt-devel] Build instructions for regular OpenWRT with Ceropackages

Sebastian Moeller moeller0 at gmx.de
Wed Jul 1 07:55:57 EDT 2015


On Jul 1, 2015, at 12:32 , Mikael Abrahamsson <swmike at swm.pp.se> wrote:

> On Wed, 1 Jul 2015, David Lang wrote:
> 
>> not true, the switch doesn't give any way for traffic to get from one vlan to the other one, so if you have gig-e connections on both sides, the traffic going from one to the other will have to go through the soc, so if there is more than 1Gb of traffic in either direction the interface will be saturated.
> 
> I don't see how it can be.

	If “saturated” means running at line rate, then I agree with David.

> The switch has a 4 externally facing ports, these all go to a single SoC port that is GigE, so the SoC cannot ingest more than 1G of traffic from the 4 LAN ports.

	But this is not SoC specific, no? If the router only had two ethernet ports and the LAN aggregation would happen at a dedicated switch between the clients and the router’s lan port the situation would be the same?

> The L2 switch chip will do the egress dropping from LAN ports->SoC ports, so there is no AQM there.

	Ingress/egress are always relative to something I guess. You can put AQM on any of the two SoC ports (or both) but that will not affect how the switch copes with overloads, unless the switch grows AQM (as Dave mentioned in earlier threads). I would hope that the LAN’s short RTTs and high bandwidth should make this bearable.

> 
>> The problem is if you have a slower connection, the bottleneck is in the switch not the soc. you may be able to set the soc-switch interface to 100Mb (make sure you have access through another interface in case you cut yourself off) and that would make the soc see the queue directly.
> 
> That is my point. There is no way by doing traffic LAN<->WAN to get egress congestion on the SoC ports, and it's on the SOC ports we can do AQM.

	Unless you reduce the egress rate to below the ingress rate. This really is why we need the soft shapers at all, otherwise the buffering moves from where we have control back into the DSLAMs/CMTSs/BRASs and CPE.

> 
> The SoC ports is gigabit ethernet only, no 10/100 available according to ethtool.
> 
> So the only way to generate congestion egress on the WAN SOC port is to add traffic locally from the SoC (iperf3 for instance), or adding traffic from Wifi.

	Well, that os shaping the other direction, so that more data can enter the router than leave. It is all a balancing game in the end…

Best Regards
	Sebastian


> 
> -- 
> Mikael Abrahamsson    email: swmike at swm.pp.se
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel




More information about the Cerowrt-devel mailing list