From: rick.jones2@hp.com   
      
   lancer6238@yahoo.com  wrote:   
   > Thank you Noob and Rick, the data packet with FIN bit set should be   
   > a piggyback FIN. I learnt something new today.   
      
   In the heyday of web server benchmarking some effort was expended to   
   ensure that the web server's TCP's FIN was piggybacked - things like   
   special flags to "senfile" calls to say "and please also initiate   
   connection shutdown" and such.  The reason being that particularly   
   with ChecKsum Offload (CKO) and copy avoidance (eg sendfile), sending   
   a standalone FIN segment was as many CPU cycles to send as a data   
   segment.  And the quantity of data returned in the mythical "average"   
   URL was about 10 segment's worth, so avoiding an additional segment   
   there was worthwhile.   
      
   Today, with on average larger (guessing) URLs, the extra steps taken   
   to ensure the FIN was piggybacked might not have happened - saving one   
   segment out of say 64 isn't quite the same ROI. Perhaps.  TSO and   
   system call overhead comes into play I suspect. And there is a greater   
   chance of a "natural" piggybacking of the FIN anyway.   
      
   rick jones   
   --   
   the road to hell is paved with business decisions...   
   these opinions are mine, all mine; HP might not want them anyway... :)   
   feel free to post, OR email to rick.jones2 in hp.com  but NOT BOTH...   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)