RE: iSCSI Autosense Consensus, Connection next steps

[Joshua Tseng <jtseng@NishanSystems.com>]

Julian,

>- the simple backpresure mechanism and TCP windows and send buffers
>proportional to the pipe rate will get you the skew down to acceptable
>levels. 

If this is the case, does the command windowing mechanism really buy
us very much?

>The total buffer set needed is still proportional
>to the total peak rate. And you will need similar mechanisms to regulate an
>asymmetric flow too
>(a bit simpler though!).

I agree.  Buffer requirements are proportional to aggregate throughput.

Josh