Firstly, thank you to ttgapers everyone else who has built this. I have finally gotten around to installing and given that I already had good ping & jitter results (DSLReports AAA) I was surprised and very please that these are now A+A+A+.
In case this is of help to anyone I have noted my setup and what has worked best - using the betterspeedtest.sh script to test.
I have an FTTC connection (VDSL2) where the speeds are capped by the ISP at 80/20 (max speed according to the modem are 96/25)
Before starting and testing with minimal traffic I was getting
Following the notes and looking at the Cake pages I set my speeds as 85% of notional speed / 95%of speedtest speeds (i tried various speedtests), and I ended up using 68/17 - when I have some time I will see if I can increase these without causing any issues.
I tested while running multiple bittorrent uploads and downloads through a VPN on the router and multiple YouTube videos on another machine (no VPN).
I then tried addding
ack-filter - better,
ptm & pppoe-ptm - also better but neither better than the other
dual-scrhost on uploads - worse
dual-dsthost on downloads - worse
diffserv4: no perceptible difference
metro - much better (as I am in a large city I though this would be worth trying)
Despite running the post installation tests under load the figures were noticeably improved.
All of the above - very satisfactory.
The trade off is the reduction in speed - as noted above, my line speed is 80/20 and tested speed without cake (again using betterspeedtest.sh) is 74.3/18.6). Losing the hardware acceleration means with cake enabled the speed drops to 52.0/13.7, a nearly 30% drop in both directions.
In case this is of help to anyone I have noted my setup and what has worked best - using the betterspeedtest.sh script to test.
I have an FTTC connection (VDSL2) where the speeds are capped by the ISP at 80/20 (max speed according to the modem are 96/25)
Before starting and testing with minimal traffic I was getting
Code:
Latency: (in msec, 61 pings, 0.00% packet loss)
Min: 5.697
10pct: 7.540
Median: 11.673
Avg: 11.945
90pct: 15.455
Max: 23.222
.............................................................
Latency: (in msec, 61 pings, 0.00% packet loss)
Min: 5.562
10pct: 19.184
Median: 25.683
Avg: 25.419
90pct: 31.732
Max: 33.907Following the notes and looking at the Cake pages I set my speeds as 85% of notional speed / 95%of speedtest speeds (i tried various speedtests), and I ended up using 68/17 - when I have some time I will see if I can increase these without causing any issues.
I tested while running multiple bittorrent uploads and downloads through a VPN on the router and multiple YouTube videos on another machine (no VPN).
I then tried addding
ack-filter - better,
ptm & pppoe-ptm - also better but neither better than the other
dual-scrhost on uploads - worse
dual-dsthost on downloads - worse
diffserv4: no perceptible difference
metro - much better (as I am in a large city I though this would be worth trying)
Despite running the post installation tests under load the figures were noticeably improved.
Code:
Latency: (in msec, 61 pings, 0.00% packet loss)
Min: 5.702
10pct: 5.867
Median: 6.648
Avg: 6.687
90pct: 7.404
Max: 8.881
.............................................................
Latency: (in msec, 61 pings, 0.00% packet loss)
Min: 5.430
10pct: 5.798
Median: 6.642
Avg: 6.733
90pct: 7.611
Max: 9.091All of the above - very satisfactory.
The trade off is the reduction in speed - as noted above, my line speed is 80/20 and tested speed without cake (again using betterspeedtest.sh) is 74.3/18.6). Losing the hardware acceleration means with cake enabled the speed drops to 52.0/13.7, a nearly 30% drop in both directions.
Last edited:
