> We have written a benchmark/test tool which attempts to emulate QT Players
> better than RTSPAM did. We notice that RTSPAM does several things
> differently that a real QT Player (QTP) which changes Server load when
> benchmarking.
> a). RTSPAM issues an "OPTIONS" RTSP command not present with QTP.
> b). RTSPAM specifies "Range: npt=0.000000-", while QTP specifies both
> starting and ending npt time.
> c). RTSPAM sends RTCP Receiver Reports (type 201) and RTCP App defined
> (type 204) packets with all zeros and omits SDES (type 201); QTP sends
> meaningful data for all these RTCP structs.
> d). The QTP RTCP App defined packet dynamically changes its length, while
> RTSPAMs is constant (and useless).
>
> Could anyone explain why these differences exist, or estimate how they
> could skew Server performance from using RTSPAM compared to actual QTPs?
> Is the QTP App-define RR only useful for "reliable UDP"? Could we have
> access to the QTP algorithm for purposes of returning more meaningful RRs
> in our test tool? (We have no interest in decoding the RTP streams).
>
> Thanks,
> Don
