[ns] goodput calculation

["pawan k gupta" <pawan.g@lycos.com>]

hi all,

thanks for the reply. but i am still confused regarding goodput calculation.

is this definition of goodput correct.

goodput = (max no of pkts recvd by the rx in sequence)
          / (total number of pkts sent by the sender including retransmissions)

i was thinking that first parameter can be calculated if we know ack_ and second parameter is ndatapack_.
of course assuming here that all pkts are of same size.
also assuming here that once rx has got pkts in sequence it will pass on the pkts to application layer.

taking an example: if current ack_ is 11 and the pkts sent by the sender are
1, 2, 3, 4, 5, 6, 7, 8, 6, 7, 8, 9, 10
(here 6, 7, 8 are transmitted twice )

so the goodput is 10/13.

thanks and regards
pawan

>to try and calculate goodput, isn't it enough to just find out the
> largest packet number that was acked by the tcp receiver and
> divide it by the total number of packets transmitted by the
> sender.
>i found that two such parameters already exists in the ns. these are ndatapack_ and ack_. 
according to this goodput of the connection is 

>(ack_ - 1)/ndatapack_ 

>please correct me if i am wrong. 
>No. (and that really messes up with SACK).

>That seems to be aiming for a ratio of effective throughput to packets
sent, or goodput to initial throughput, if you like.

>Also, in TCP, packets can be variable-sized (anything less than path
MTU), although ns is a bit skimpy in simulating that.

>If you want goodput, you really have to track the left edge of the
window at the receiver as it passes data to the application - and
ndatapack_ isn't at the receiver.



Get 250 color business cards for FREE!
http://businesscards.lycos.com/vp/fastpath/