?

Log in

No account? Create an account
Some more hacking and neonlights is starting to do something useful.… - Artur Bergman [entries|archive|friends|userinfo]
Artur Bergman

[ website | O'Reilly Radar ]
[ userinfo | livejournal userinfo ]
[ archive | journal archive ]

[Aug. 30th, 2007|07:58 pm]
Artur Bergman
[Tags|]

Some more hacking and neonlights is starting to do something useful.

The packet loss is actually retried packets, which is subtly and sadly not the same. In fact, since this dump was run on the destination, we can assume that retries from src is actually acks getting lost from the destination to the source, which would be packet loss from dst->src and not as indicated src->dst.

It currently does around 1000 packets per second on my laptop. By rewriting the core in C, I hope to avoid the marshal cost.

Code is at http://neonlights.googlecode.com/svn/trunk/lib/Neonlights.pm and is even commented, but undocumented.

Streams: : 3268
Total Packets : 101012
Total Retries : 6881
Packet loss % : 6.81
Avg packets/stream: 30
Avg retries/stream: 2
***** FROM src ****
Total Packets : 59976
Total Retries : 6016
Packet Loss : 10.03%
Avg packets/stream: 18
Avg retries/stream: 1
***** FROM dst ****
Total Packets : 41036
Total Retries : 865
Packet Loss : 2.11%
Avg packets/stream: 12
Avg retries/stream: 0
LinkReply