I want my router to make upstream packet to goes first from lan ip 192.168.1.100 and i want downstream packets to have the lowest priority how can i achieve that???
For prioritizing upstream packets you can install Qos package and add the IP for express forwarding.
For changing priority to downstream packets you need to ask from your ISP.
What is the type and speed of your WAN link?
Which application or use case are you trying to optimize?
pppoe and speed upload 2mbit and download 30mbit i am trying to optmize gaming apps on ps4
I am using vdsl
VDSL uplink and downlink work in parallel (full-duplex), there should be no need to prioritize one over the other.
I would suggest to set up SQM to reduce the latencies, and measure the results as described in the wiki article.
30/2 MBit/s is not very demanding on the router CPU, as long as no VPN is involved.
I'll ask anyway, just in case: Which model is your router?
my modem is zte ZXHN H168N V3.1 and my router is pc ( openwrt x86)
and these are my vdsl status:
Link Status Up
Modulation Type VDSL2
Actual Rate(Up/Down) 5005/39999 kbps
Attainable Rate(Up/Down) 30628/138444 kbps
Noise Margin(Up/Down) 37.8/20.2 dB
Line Attenuation(Up/Down) 0.3/8.2 dB
Output Power(Up/Down) 4.1/13.6 dBm
Data Path(Up/Down) Interleaved/Interleaved
Interleave Depth(Up/Down) 49/2531
Interleave Delay(Up/Down) 5/9 ms
INP(Up/Down) 2/8.9 symbols
Showtime Start 6 h 13 min 5 s
CRC Errors(Up/Down) 0/6
FEC Errors(Up/Down) 77/33479
Is it set up as a router, or in bridge mode?
I would prefer bridge mode, if possible, and handle PPPoE on the OpenWrt router.
Good, should be fast enough unless it is very old.
Looks like the link speed is 40/5 MBit/s.
The limitation to 30/2 may be caused by shaping on the provider side and protocol overhead.
Which results do you get from the DSLReports speed test mentioned in the SQM article?
this is my dslreport result with sqm ,sqm setup:( eth1 wan speed download 24mbit, upload 1.6mbit) ( pppoe-wan speed download 24mbit and upload 1.6mbit) ( both uses piece of cake script with overhead 26)
yes that's actually my setup modem bridging to open-wrt
The results look good, although the idle latency at 90ms is a bit high. However, this latency may be introduced on a link outside your control, you can use traceroute or mtr to find out.
This probably just reflects the indirect connections from the OP (in egypt?) to the data centers the test automatically selected. The bigger issue I see is that instead of requested 24 Mbps (24 * (1500-8-20-20) / (1500-8+26) = 22.95 Mbps speedtest goodput) only 6.7 Mbps actually wer achieved...
is there a fix to this ?
A fix to what exactly? I generally like and recommend the dslreport speedtest a lot, but it seems that even the closest "nodes" are ~90ms away from you (which is not terrible, but also not great). You could register for free at dslreports (unless you did that already) and then you can configure the set of server locations you want to allow in the speedtest, see https://forum.openwrt.org/t/sqm-qos-recommended-settings-for-the-dslreports-speedtest-bufferbloat-testing/2803 for recommendations how to set up the dslreports speedtest. But I feat the closest servers will be the european ones.
About not getting close to the expected goodput, this might well be related to the actual measurement, so I would very much also try the following:
speedtest.net, Ookla's speedtest networks has nodes in egypt
to see whether the low speed results might be more related to your ISPs connectivity to europe than your connectivity to your ISP.
This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.