> Hello Markus,
> Fragmentation has nothing to do with ICS, it is IP issue and you cannot
> influence it. 

Ok, understood. I just thought to have seen a flag somewhere
"don't fragment".

> I think if you send packets smaller than MTU then chance
> is lower that you get fragmentation. Also I think on a busy network
> chance is higher. But what is the problem? Wy take the risk? Just
> concatenate  received packeds in case you don't have the complete one.

Sorry, you didn't fully understand what I wrote. On the PC side I have 
no problems concatenating several chunks of received data.

On the other side I have no chance: a too long gap between 2 bytes is 
regarded as new packet and the old is discarded because the checksum 
would be wrong. This has nothing to do with TCP/IP but simply with the 
payload, means routing of another protocol over TCP.

Tests on a otherwise not used LAN showed that TCP works fine in that 
configuration, but other tests on a different (our company's real LAN) 
showed that I can cause trouble if I do a port scan on that TCP 
interface converter. I'm not sure why this is and my test app. is a bit 
simplistic written.

So rigth now I'm not sure whether I should tread that port scanning case 
as some unlikely cause or if I should say: TCP doesn't work reliably 
enough for my case and look for other technologies. Thus the question 
about fragmentation. My normal case will be a LAN, but if sales then 
decides: oh using it for support via Internet I could maybe get into 
trouble, but I sort of doubt it as my packets a < 64 byte.


To unsubscribe or change your settings for TWSocket mailing list
please goto http://lists.elists.org/cgi-bin/mailman/listinfo/twsocket
Visit our website at http://www.overbyte.be

Reply via email to