Early statistics

The latest implementation of my HPPC system provides means to add rules and match an IP-address against the system.
At this point I'm getting about 3.7 million classifications per second, but I'm not sure if this rate will hold. Some more code needs to be added to actually go through all matching rules and match against other factors (custom matches like IPP2P for example), aswell as some optimisations to speed up everything.

But 3.7 million classifications as a first estimate is not bad. Most packets will have at least 4 fields to match against: source/destination IP and source/destination port number. This could result in about 900K packet-classifications per second (or possibly 900.000 packets per second throughput). Assuming an average of 200 bytes per packet, this would give a throughput of 180MB/s which is more than 1Gb/s.


(deepstar/tachyon) ~/projects/hppc$ time ./a.out 400000000 `cat data/mccoy1-sources.txt`

real 1m53.718s
user 1m46.495s
sys 0m0.216s
(deepstar/tachyon) ~/projects/hppc$ bc
bc 1.06
Copyright 1991-1994, 1997, 1998, 2000 Free Software Foundation, Inc.
This is free software with ABSOLUTELY NO WARRANTY.
For details type `warranty'.
400000000/106
3773584
(deepstar/tachyon) ~/projects/hppc$