Re: [OPENETLAB] Question: Oplets and Accelar Queue Mgt./Routing

[Rob Jaeger <rfj@cs.umd.edu>]

On Thu, 21 Sep 2000, Phil Wang wrote:

> "Sawant, Vivek" wrote:
> 
> ... ...
> > >
> > > RED implementation would have to be hardcoded, however ability to pass
> > > parameters like max threshold, min threshold, burst, probability of drop
> > > using a mapping from JFWD is probably prefered.
> > 
> > That's fine. I am trying to see if one can use ORE/JFWD set up as an
> > experimental platform for trying out new queue management strategies.
> > I don't mind if the RED is absent or not activated in Accelar.
> > If I can send packets to Accelar (output ports), I would be happy.
> Did you mean that the incoming packets matching particular filters
> (using JCapture) are diverted to the CPU, and an Oplet service
> (implementing RED or alike) queues then forwarded them to the Accelar
> egree ports?
> 
> The answer is yes! However, the traffic must be experiemental since the
> Accelar does not have enough memory to buffer those packets.

won't the source IP address be (incorrectly) the IP address of the router 
performing the RED operation??