Hi, On Thu, Feb 21, 2019 at 01:19:57PM -0800, Mike Wright wrote: > I followed "Using netem on one..." with interest and want to thank > the participants. I'd never come across those terms (strong host, > weak host, arp flux) before. Researching them broadened my > networking understanding tremendously. > > The following link hit all the bases in your discussion: > https://unix.stackexchange.com/questions/258810/ > linux-source-routing-strong-end-system-model-strong-host-model > ?answertab=votes#tab-top Thanks for the link. > My question has to do with cost between the models where cost refers > to time to route. Since the strong model seems to involve a lot > more decision making is there an appreciable difference in the time > spent routing versus the weak model or is the difference so small as > to be inconsequential? I would not expect that the choice of strong or weak model to have an effect on "routing" performance. Simple checks suffice for the strong ES model: discard if ingress interface IP is different from destination IP, outgoing interface is that with the same IP as the packet to send. I would not expect significant difference in performance, and the strong ES model might even help performance a tiny bit. If one thinks about a gateway with dedicated forwarding hardware using TCAM, e.g. a switch, checking if a packet is allowed to ingress or egress an interface has no impact on forwarding performance. Thanks, Erik -- The Linux approach is to do the job right. [...] This is the Linux way of doing things. It's like the GPL and being shouted at by Linus. They are things you accept when you choose to take part. -- Alan Cox
