Crazy long title, but current situation creates complex problems and needs good/smart solutions
Unfortunately with families at home and everybody doing home working in the city my LTE connection speed varies heavily over the day:
In the morning LTE gives (for me) great ~20-30Mb/sec DL and 12-20Mbps UL throughput
But over the day this goes down and in the later afternoon/evening specially the UL is just a slim 1-2Mbps (while DL is still around 4-8Mbps
My DSL line is (as you expect) stable all day around 5M down and 700k up (guys stop laughing , that is all I can get here )
I use PCC which is static setup (and queues as well for each WAN access, DSL and LTE).
If I fit this for the āmorning use caseā its completely spoiled in the afternoon/evening. I could do a schedule and make 2 sets but its not smart.
In addition the queues always use absolute values (its a pitty you can not use a variable or percentage there). So I would also
need to create 2 sets of Queues for my 6 LAN subnets. This I really want to avoid.
What could one do to balance this out or adapt the PCC rules and queues in a more āsmartā way? Or is there a different way of āload balancingā that can easier be adapted
to changing environment?
As you see from my speed limits, I really need all I can get with whole family being at home and on the computers the whole dayā¦ please HELP
In crazy days do a crazy workaround with PoE to get better LTE signal like
(photo from shared, sharedā¦ FB group and I no possibility to found a source).
About QoSā¦ you know that is a problematic in Dynamic WAN interface like lte1.
I have the same situation but my workaround with move my SXTR to ābetter placeā solve me a problem for now.
maybe I should install my Audience on a turn table, that always looks for the best connection speed
Today UL is stable (but low):
This morning I got this speed:
Then this afternoon I get this speed:
The other day it was even worse with Ā±4Mb DL and 1-2Mb uplink (while 2 months ago I always had UL over 20Mb!!!).
With such variations its really hard to make good bandwidth control, and shows the limit of LTE routers for home usageā¦
How is your device comparing to LHG CAT6 (I also have this but did not want to do a fixed install at home of thatā¦). Yours looks
smaller and less āsat dishā than the LHGā¦
If you had 20 Mbit/s with your Audience standing at the same spot in your living room, the LTE issue is clearly not the signal quality but the number of other users on the same BTS (or, worse, same backbone). So speaking highly theoretically, a directional antenna could allow you to connect to a less occupied BTS, e. g. if you live close to an office-only area where no one uses LTE these days.
Using another SIM in another LTE device would allow you to have two slices of the cake which is equally split among all the users of the same BTS.
Regarding the dynamic adaptation of the traffic distribution rules to the changing load of the LTE network, the first question is how to determine the currently available bandwidth. As most of the time you are saturating it, it is relatively easy to use dedicated raw rules to count downloaded bytes - every 30 seconds or so, you can read the values and reset the counters, calculate some averages so that you wouldnāt overreact to short-time glitches, and adjust your load distribution rules accordingly. But if you throttle the delivery of the downloaded data to clients, and there is no client which has no limitation, you might get trapped into a self-locking loop. So people here have recommended to ping something outside and watch the response time - if the ISPās queues in the download direction are clogged with other data, the ping response will be longer than when the queues are empty. So if responses are quick, you can add traffic to the LTE.
But Iām afraid it may be quite a lot of work to tune this into something practically useful. Of course you have to move from choosing the WAN using PCC to choosing it using random (with connection tracking), as with a random rule, you can change the load distribution ratio between the WANs by changing a single value in a single rule in range 1-99, whereas with PCC the modification of the distribution ratio requires more rules and changes.
indeed I did some outdoor tests with the LHG CAT6 I have, but did not found much better performance (tests were done before the COVID times).
It was connecting to the same cells / base station as the audience, even if I pointed it to other towers (I looked up their location on the internet).
Thus I did not want to take all the hassle to install an outdoor antenna, wiring etc. and stayed with the audience.
If I have time and SIM card will do some tests in the coming days with that device again, maybe this time its working betterā¦
All this would be less of an issue if you could use global variables in the QUEUE TREE parent queue for the limit, and then use
percentages of parent queue in the child queue. It would allow to increase/decrease your whole bandwidth management smoothly.
As you canāt do that you need do use absolute values.
using Mikrotik bandwidth test result and then save the outcome as global variable, use it to adjust your routing / queue
downside is speed test along with live internet traffic is not accurate, I donāt think you want to stop all traffic just to do a speed test, you can but probably not at the cost of rest of the family shouting at youā¦
I would look into swapping SIM or get a 2nd SIM, do your research and see which network operator is less likely to suffer from drastic speed reduction during the day in your area.
)
