Hi
I posted this a week ago on misc without any result. Waited a few more days
before I posted it here. And I hope the "pf" list is a better forum than misc
for this type of question...
As the fact is that the only way to reduce the worst case delay is to increase
the bandwidth reservation I have the following question.
(above statement from a technical overview of HFSC)
Let's say that I set an initial realtime bandwith for 1 second and then a
lower value (example: realtime 1Mb 1000 0.5Mb). Then I assume I will have a
lower delay for the first second as the bandwidth is higher...
Now... What happens with the initial delay if I have for example:
(only a paper constructed example that is not tested)
altq on $EXT hfsc bandwidth 10Mb qlimit 100 queue {clientnets, std } queue
clientnets bandwidth 1% qlimit 100 hfsc (realtime 1500Kb, linkshare 7500Kb,
upperlimit 8500Kb) { XXX, YYY, ZZZ }
queue XXX bandwidth 1% qlimit 100 hfsc (realtime (1000Kb 1000 500Kb),
linkshare 2500Kb, upperlimit 6000Kb)
queue YYY bandwidth 1% qlimit 100 hfsc (realtime (1000Kb 1000 500Kb),
linkshare 2500Kb, upperlimit 6000Kb)
queue ZZZ bandwidth 1% qlimit 100 hfsc (realtime (1000Kb 1000 500Kb),
linkshare 2500Kb, upperlimit 6000Kb)
queue std bandwidth 1% qlimit 100 hfsc (realtime 500Kb, linkshare 500Kb,
upperlimit 100% default ecn)
and the average load on the XXX, YYY and ZZZ queues are aprox 2500Kbit?
Will I have any benefit at all of the higher initial realtime value during the
first second if the average load is always higher than the realtime values? Or
could I set realtime to 0 without any drawbacks?
Thanks in advance
Per-Olov Sjöholm
--
GPG keyID: 4DB283CE
GPG fingerprint: 45E8 3D0E DE05 B714 D549 45BC CFB4 BBE9 4DB2 83CE
