DVTS has bursty traffic, since it does only frame discarding. When a undiscarded frame is sent, it causes bursty traffic. To solve the burstyness of the DVTS traffic, use with ALTQ is recommended.
DVTS-0.1.? can send video and audio in different streams. When video and audio are sent in different streams, the stream for audio is sent via port number [video port number + 1]. When video and audio are send separately, shaping them can result to better quality of DVTS.
ALTQ can be used as a shaper for DVTS. But, when using ALTQ-2.0 for DV, it is recommended to change the following part. This is because DV stream sends a lot of packets within short terms.
libaltq/qop_hfsc.c:144: int qlimit = 50;
to
libaltq/qop_hfsc.c:144: int qlimit = 500;
For example, the configuration using H-FSC of ALTQ can be like the following. The following is the configuration I used when sending 1/30 DV stream with video and audio in different streams.
#
# for DV at 10Mbps NIC
#
interface fxp0 bandwidth 10M hfsc
#
# 100% of the excess bandwidth goes to the default class
#
class hfsc fxp0 def_class root pshare 15 default
#
# DV audio 3Mbps guaranteed rate (no excess bandwidth assigned)
#
class hfsc fxp0 audio_class root grate 3M
filter fxp0 audio_class 0 1396 0 0 17
#
# DV video 2Mbps guaranteed rate (no excess bandwidth assigned)
#
class hfsc fxp0 video_class root grate 2M
filter fxp0 video_class 0 1394 0 0 17
The command I used for this at the sender, was,
% dvsend6 -s 15 -h desthostname -f 30 -d