The official Mikrotik router thread

75Gb of test data later, I got it to this:
View attachment 1761717
Using these on my 500/500 line:
View attachment 1761718
So ran some tests with my son over the weekend.
Historically, when he downloads and I am playing PUBG on the PS, the ping remained the same, but packet loss (up and down) jumps between 10% to 25%.
After the changes, there is in increase in ping, from around 170ms to 210ms-ish, but no packet loss.
 
So ran some tests with my son over the weekend.
Historically, when he downloads and I am playing PUBG on the PS, the ping remained the same, but packet loss (up and down) jumps between 10% to 25%.
After the changes, there is in increase in ping, from around 170ms to 210ms-ish, but no packet loss.
Get him to run some more tests while you play.

It's a win-win situation. He gets to download, you get to play PUBG and optimise your network :)
 
Get him to run some more tests while you play.

It's a win-win situation. He gets to download, you get to play PUBG and optimise your network :)
He is back at varsity, so not at home :)
I am happy with the ping increase rather than packet loss. In a BR with already higher than your opponents' ping, packet loss on top of that literally kills you :)
 
For sure, but if you could keep ping more or less the same, that would be really ideal.

I absolutely LOVE the mangle feature on the Mikrotiks, but it does tend to hammer the CPU a bit :X3:

I apply packet marking and then use those to prioritise my traffic into eight queues, which is working out really well. No complaints in the household yet.

1728372651569.png
 
For sure, but if you could keep ping more or less the same, that would be really ideal.

I absolutely LOVE the mangle feature on the Mikrotiks, but it does tend to hammer the CPU a bit :X3:

I apply packet marking and then use those to prioritise my traffic into eight queues, which is working out really well. No complaints in the household yet.

View attachment 1762872
I both respect this and cringe at it.
I am not sure why you are marking the packets of so many individual clients, but I can respect the efforts taken here.
You appear to be a man who has never lost a 10mm socket in his life.
 
I both respect this and cringe at it.
I am not sure why you are marking the packets of so many individual clients, but I can respect the efforts taken here.
You appear to be a man who has never lost a 10mm socket in his life.
This is what happens when you have a kid that won't sleep and then proceeds to cause you to also not be able to fall back asleep. 3-4AM is peak network optimisation time for me :ROFL:

My network is basically static and not much changes. I found that vendors don't bother marking their packets with DSCP properly, so I just tagged my most frequent clients (assigned them all static IP's) and the rest is picked up by the 'scavenger' queue that assigns everything else whatever else is left. I just hid the actual packet mark column as it had some personal info :)
 
For sure, but if you could keep ping more or less the same, that would be really ideal.

I absolutely LOVE the mangle feature on the Mikrotiks, but it does tend to hammer the CPU a bit :X3:

I apply packet marking and then use those to prioritise my traffic into eight queues, which is working out really well. No complaints in the household yet.

View attachment 1762872
Yep, that is the dream for sure ... but I don't know where to from here, to get the ping similar. Cause the bloatware test shows minimal impact on ping, but it appears international takes a bigger hit
 
This is what happens when you have a kid that won't sleep and then proceeds to cause you to also not be able to fall back asleep. 3-4AM is peak network optimisation time for me :ROFL:

My network is basically static and not much changes. I found that vendors don't bother marking their packets with DSCP properly, so I just tagged my most frequent clients (assigned them all static IP's) and the rest is picked up by the 'scavenger' queue that assigns everything else whatever else is left. I just hid the actual packet mark column as it had some personal info :)
LOL, I don't think I have ever heard someone say that a packet mark contains personal info...

1728385167163.png
 
Last edited:
LOL, I don't think I have ever say that a packet mark contains personal info...
Haha! Nah, nothing sinister (or kinky lol), I just categorise some packets according to the device, and some of them are named with me or my wife's names in there so I don't get confused.

1728387183691.png
 
Haha! Nah, nothing sinister (or kinky lol), I just categorise some packets according to the device, and some of them are named with me or my wife's names in there so I don't get confused.

View attachment 1762954
WiFi connected sex toys... we need to prioritise those packets for sure!
 
Yep, that is the dream for sure ... but I don't know where to from here, to get the ping similar. Cause the bloatware test shows minimal impact on ping, but it appears international takes a bigger hit
I think you might need to play around with maybe placing your PS in your queue tree as a top priority device, and set the 'limit at' (MT speak for guaranteed bandwidth) at maybe 7500k or 10m and see if that improves things.

Otherwise, assign a static IP to your son's downloading device and then using queues, you can give him a 'lower' (higher, in this case) prio which should also help. You might need to also play around with different queue types. Unfortunately there is no one size fits all, but figuring it out is half the fun.
 
This is what my queue setup looks like:

1728412917802.png

1_frogfoot is the parent queue with default settings
udp is my DNS and icmp traffic for Adguard Home and Smokeping
fnb is my work-related stuff
tv is my Nova box and the Hisense TV's IP addresses
phireside and phiresidewife is our devices - phones, laptops, and kindles
tinyplex is my two wired machines, a Lenovo Tiny and a Dell Optiplex (see how clever I was with the naming? Huh?)
iot is a separate vlan for the iot goodies. Light switches, dimmers, smart bulbs, etc.
scavenger is everything else that does not fit into the above categories.

I use Mangle rules to mark packets from the respective endpoint IP's and so far it works pretty well. But I am always tweaking and changing stuff so in a week from now it will likely differ again :) . I've experimented with different upload and download limits, burst settings etc. but 55M seems to work best on my 60/60 Vrotvoet FTTH link.
 
Top
Sign up to the MyBroadband newsletter
X