home bbs files messages ]

Forums before death by AOL, social media and spammers... "We can't have nice things"

   comp.protocols.tcp-ip      TCP and IP network protocols.      14,669 messages   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]

   Message 13,606 of 14,669   
   jack to Mark Hobley   
   Re: Path Maximum Transmission Unit Disco   
   26 Sep 10 16:20:45   
   
   XPost: comp.os.linux.networking   
   From: jcfmasters@yahoo.com   
      
   Mark Hobley wrote:   
   > On Sun, 26 Sep 2010 12:23:58 +0200, Ersek, Laszlo wrote:   
   >   
   >> The discovery and fragmentation work in the IP layer. The HTTP proxy   
   >> works in the TCP (and application) layers. Once Machine B is reached,   
   >> the IP layer's job is done. That job recommences towards Machine C.   
   >   
   >> Machine A and Machine B are constant links in the above chain. Machine B   
   >> connects to lots of different HTTP servers (or further proxies) scattered   
   >> all over the net. Machine C is variable, and thus each type of B-C link is   
   >> different. Therefore it makes sense for the discovered A-B and B-C PMTU's   
   >> to differ and to have different lifetimes.   
   >   
   > If Machine B is receiving data from Machine A in 1500 byte segments, and   
   > then retransmitting this in smaller segments to suit C, then PMTUD has not   
   > really prevented fragmentation. Wouldn't it have been better to just do away   
   > with PMTUD and let the fragmentation and reassembly take place downstream   
   > towards C?   
      
   PMTUD between A and C is not applicable in this case. If B functions as   
   a proxy, there is a TCP connection between A and B, and a *different*   
   TCP connection between B and C. PMTUD only applies to a single TCP   
   connection.   
   For all you know, there might not even /be/ a connection between B and   
   C, and B is serving up a file from its cache.   
      
   >   
   > What benefit do we get from Machine B knowing the Maximum Segment Size for   
   > each of the target hosts?   
      
   None, if B is connected directly to C. Plenty, in the (somewhat extreme)   
   case that B connects to C via routers R1, R2, and R3 with respective   
   path MTUs of 1480, 1470 and 1450. B sends a 1500-byte packet; R1 splits   
   it into a (1480+20). R2 splits that into (1470+10+20). R3 splits that   
   into (1450+20+10+20). Now we have 4 packets, with the overhead of 3   
   extra headers, that need to be re-assembled at C. And without PMTUD this   
   will happen for every single packet larger than 1480.   
      
   -j   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]


(c) 1994,  bbs@darkrealms.ca