Hi to all, I want to use HFSC scheduling at output interfaces of software routers in my testbed. I used HTB before, but since that is not good solution for delay sensitive applications I decided to try HFSC. For the beginning, I connected two host to one PC router: H1(8.8.8.12)----Router-----H2(8.8.8.11) However with HFSC I got much worse results then with HTB. I cannot figure out what I did wrong. My experiment was really simple. From H1 I generated one UDP flow with rate 100kbit/s to H2. On output interface of the router (towards H2) I configured policy in the following way: tc qdisc add dev eth0 root handle 1: hfsc default 11 tc class add dev eth0 parent 1: classid 1:1 hfsc ls rate 100000kbit ul rate 100000kbit tc class add dev eth0 parent 1:1 classid 1:11 hfsc rt m1 400kbit d 30ms m2 100kbit tc class add dev eth0 parent 1:1 classid 1:10 hfsc ls rate 10000kbit ul rate 10000kbit and defined tc filter: tc filter add dev eth0 protocol ip parent 1:0 prio 1 u32 match ip src 8.8.8.12 flowid 1:11 I synchronized hosts' clock and run tcpdump on both to measure delay. Results were disappointing because some of the packets experienced delay of almost 200ms instead od 30ms. I have never had such large delay with HTB. The longest delay I had with HTB was about 150 ms. Can you please help me to solve this problem? I do not see any other Linux scheduling alternative that can guarantee delay... Regards, S. -- To unsubscribe from this list: send the line "unsubscribe lartc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
