Jimmy Fox writes: We don't have bad latency. Just huge amounts of dropped packets.
Sorry, I wrote "latency" when I meant "packet loss" as measured by utcapture. We can easily trigger 20% packet loss on simple flash animations. At this level, the desktop is largely unresponsive. Darrel makes me think that perhaps because all DTU traffic is funneled through a single interface the server doesn't know how to adjust for the lower network's bandwidth. Since the local LAN is 100Mb, it doesn't know to throttle down for the 3Mb network? I only have one Interface to provide to the DTU traffic so I can't test this theory. Well...we have exceptionally good networking. We tested 100 Mb/s due to recommendations by Sun for somewhat similar problems. In our case, connecting to 100 Mb/s Cisco switch ports helps Ray 2 packet loss, but has little affect on the Ray 1 problem. Our experience where we have very little packet loss on direct connections through inexpensive switches, or routing through another host with Rays on one interface and the world on the other, should explain something. We finally admitted defeat and used workarounds (either a host with two interfaces doing the routing, or replace with Ray 2 and run the X4200 interface to a 100 Mb/s Cisco switch port). -- Darrel Hankerson _______________________________________________ SunRay-Users mailing list [email protected] http://www.filibeto.org/mailman/listinfo/sunray-users
