White paper performance numbers

[nnmk]

If this is true: 2060 x 700 Kbps ~= 1.4Gbps
=> Then the claim that the results published in the FMS3 White Paper are
limited to 1Gbps throughput
([url]http://www.adobe.com/products/flashmediaserver/pdfs/FlashMediaServer3_WhitePape[/url]
r_ue.pdf , page 17) can not be true!

Can someone comment this please, maybe i am not understanding the issue
correctly BUT the numbers published are very confusing (pages 16, 17), the
complete benchmark for the FMS 3 would be a benefit to us all...

[techeye]

There is reference to 2 bonded nics (why? I have no idea....), but the result
of ~1.4 Gbps is consistent with my bonded nic tests on rhel4 (2.6.29-ELsmp).
Fyi, the results with the newer rhel5 kernel (2.6.18) was able to actually get
up to ~1.9 Gbps. The newer Linux kernel (& tcp optimizations) make a big
difference. I don't know about the results on windows.

But yeah, I agree, apparently whomever was editing this was unable to portray
exactly what they meant. They probably wanted to target the average customer,
who does not usually use bonded nics (yet). Anyway, the take away, I think, is
that performance numbers improved greatly, and that in a typical 1 Gbps
configuration one is able to fully take advantage of the entire nic (minus the
OS's tcp stack), and in fact that FMS can go even beyond 1 Gbps by utilizing a
second nic.