White paper performance numbers
if true: 2060 x 700 kbps ~= 1.4gbps
=> claim results published in fms3 white paper limited 1gbps throughput ( http://www.adobe.com/products/flashmediaserver/pdfs/flashmediaserver3_whitepaper_ue.pdf , page 17) can not true!
can comment please, maybe not understanding issue correctly numbers published confusing (pages 16, 17), complete benchmark fms 3 benefit all...
=> claim results published in fms3 white paper limited 1gbps throughput ( http://www.adobe.com/products/flashmediaserver/pdfs/flashmediaserver3_whitepaper_ue.pdf , page 17) can not true!
can comment please, maybe not understanding issue correctly numbers published confusing (pages 16, 17), complete benchmark fms 3 benefit all...
there reference 2 bonded nics (why? have no idea....), result of ~1.4 gbps consistent bonded nic tests on rhel4 (2.6.9-22elsmp). fyi, results newer rhel5 kernel (2.6.18) able ~1.9 gbps. newer linux kernel (& tcp optimizations) make big difference. don't know results on windows.
but yeah, agree, apparently whomever editing unable portray meant. wanted target average customer, not use bonded nics (yet). anyway, take away, think, performance numbers improved greatly, , in typical 1 gbps configuration 1 able take advantage of entire nic (minus os's tcp stack), , in fact fms can go beyond 1 gbps utilizing second nic.
but yeah, agree, apparently whomever editing unable portray meant. wanted target average customer, not use bonded nics (yet). anyway, take away, think, performance numbers improved greatly, , in typical 1 gbps configuration 1 able take advantage of entire nic (minus os's tcp stack), , in fact fms can go beyond 1 gbps utilizing second nic.
More discussions in Adobe Media Server
adobe
Comments
Post a Comment