karatekid wrote on Feb 18
th, 2013 at 11:01am:
In fairness, a packet filter or similar could be causing the one timeout, but the point of my post over in official was that Tuhrd went straight for "well, 130 ms is high" instead of the timeout on the edge hop and the latency on the last two... just silly.
It was Ironclan, and it looks like he never lived in the time when 800ms latency was a good value.
I know math are hard, but it's basic math :
Light propagate at the Speed of Light ( I'm assuming light in a fiber propagates at the speed of light in vacuum here ).
there's a few ten/hundred/thousand kms between one hop and another
so distance/speed of light = latency of that link.
do that for all the links and you'll get the incompressible latency... then add the router generated latency, and you'll get the full latency.
but I'm not sure he'll understand the concept... especially has he is still mixing latency, ping, lag and bandwidth.
Now if somebody can explain me why when I configure a bridge with a L2VPN on a pseudowire and a L2VPN directly on the bridge, disable the mac learning and then start injecting 500Mb/s of 64bits frames I get on the agregation switch of the access circuit a 1Gb traffic in. ( yeah I know it sounds weird... but that's my problem of the day... This is the Internet. )
[ oh and there's no way you can find a real traffic with only 64 bytes packets, but there's fucking test equipments that are RFC2544 compliant that generates this kind of fucked up traffic just to give people like me nightmares ]