> From: [email protected] [mailto:[email protected]]
> On Behalf Of Doug Hughes
> 
> As for the firewall claming 6gbits/sec when you've only sending
> 1gbit/sec sounds quite bogus. 

For a really quick simple test, I just wrote (copied from the internet) a
python tcp server, and a python tcp client.  Connected them, sending 1 byte
across repeatedly.  Sniffed them with wireshark.

I found, by default, it was setting the PSH flag, and each 1-byte payload
became 60 bytes sent to the server, and 54 bytes ack reply.  This is on a
LAN, so there is no additional overhead caused by routing encapsulation or
encryption.

I would say, 59 + 54 bytes per packet is the *minimum* TCP overhead you
could hope to achieve.  So if you average sending 22 bytes per socket write
(at most) then you will have each [PSH] packet ~ 81 bytes and each [ACK]
packet ~54 bytes, for a total TCP data approx 135 bytes round trip, and
achieve a ratio of approx 6:1 total TCP data versus payload.  If you happen
to be on 10G ether traversing a 10G router, this would probably mean 6gbits
measured by the router, while 1gbit measured by the server or client.

_______________________________________________
Tech mailing list
[email protected]
https://lists.lopsa.org/cgi-bin/mailman/listinfo/tech
This list provided by the League of Professional System Administrators
 http://lopsa.org/

Reply via email to