morgan@JESSICA.STANFORD.EDU (09/29/88)
I heard recently about a campus that was doing AppleTalk-on-IP across a 19.2Kbps link to a branch campus 150 miles away. Have others tried such things? Can you do AppleShare across such a link, or do you run into timeout problems? Has anyone tried this across the Internet? Would anyone like to try? - RL "Bob" Morgan Networking Systems Stanford
morgan@JESSICA.STANFORD.EDU (09/29/88)
> The problem you run into is that KIP does not handle IP fragments and > the intial packets sent for AppleShare (an echo packet) is larger than > the "minimally required" mtu of 576 (often packets going across imp's > are fragment around that size). Hmm, I always thought the MTU of the ARPANet was 1008. I wonder what the MTUs of the NSFNet and the various regionals are? - RL "Bob"
eshop@saturn.ucsc.edu (Jim Warner) (10/02/88)
In article morgan@JESSICA.STANFORD.EDU writes: > > >Hmm, I always thought the MTU of the ARPANet was 1008. I wonder what >the MTUs of the NSFNet and the various regionals are? > The MTU (Maximum Transmission Unit) of your regional network (BARRNet) and the NSFNet is the Ethernet maximum. I can't speak for the other regionals, but I would be very suprised if their MTU was any less. This is fairly easy to test; Berkeley unix ping has an option that allows you to specify the size of the ICMP echo request packet up to ~2000 bytes (guaranteed to fragment on an Ethernet). You need an Ether-scope or 'tcpdump' on a Sun to see if what comes back has been fragmented. You may want to contact your campus networking services office to verify which remote sites are accessed over NSFNet. It is not always easy to tell. --jim