From: barmar@alum.mit.edu   
      
   In article ,   
    keith  wrote:   
      
   > Why is it that many applications send out UDP packets of size 536? Some   
   > people recommend 1472 (1472 + 8 + 20 = 1500, the usual MTU). Judging   
   > from an old post by Rich Stevens, the most portable packet size ought to   
   > be (572 - 20 - 8 = 544)... How to choose the packet size and why the 536   
   > magic number? Is it the fault of TCP and people just copied the usual   
   > MSS? What are your experiences and what packet size do you use in your   
   > applications?   
      
   It's because of this requirement in RFC 1122:   
      
         3.3.2  Reassembly   
      
            The IP layer MUST implement reassembly of IP datagrams.   
      
            We designate the largest datagram size that can be reassembled   
            by EMTU_R ("Effective MTU to receive"); this is sometimes   
            called the "reassembly buffer size".  EMTU_R MUST be greater   
            than or equal to 576, SHOULD be either configurable or   
            indefinite, and SHOULD be greater than or equal to the MTU of   
            the connected network(s).   
      
   So if you keep your datagram under 576, it's guaranteed that the   
   receiver will be able to reassemble it if fragmentation is necessary.   
      
   --   
   Barry Margolin, barmar@alum.mit.edu   
   Arlington, MA   
   *** PLEASE don't copy me on replies, I'll read them in the group ***   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)