HI,
first .. sorry for my english.
second... i have the same problem .. i'll show my screenshot .
I have try to set my ap at 40mhz, but my bw is ever max 35mbit (with 20tcp session ... with 2 tcp session i have only about 20 mbit )
test is done from a pppoe server > ptp to the bts > mikrotik switch > ap cambium > epmp force 200 > mikrotik ap (for btest)
how u can see from the screen, the ap have only 40mbit of bandwidth in use ( 40sm registred ) .
Ap is in flexible mode and 5ms frame
i'll respond in this thread, i opsed earlier
firmware version is 2.6.1 to AP and to SM
Mikrotik is :
pppoe server a ccr 32
ptp is width netmetal and btest result is 90/80mbit so is good
to the bts there's a switch 1100ahx2, so more powerfull.
CPU problem there isn't cause not goes over 30% during the test to the SM
are you able to test each hop of the link indiviually?
if you pulled that test from 100 meg port, and the backhaul has the capcity for you to hit 95 mbps, I'd expect to see that every single hit. hitting 80 meg would be a red flag to me, unless thats the end f the BH bandwidth as you would expect it to be.
if you can test just the cambium radio pair, and you get the results you expect, bring the switch back into the equation and test again, continue doing that until you've worked your way back to your current head end, you'll likely run into the drop along the way but it may not be at fault of any device, it could simply be congestion that needs cleared out or handled different in a different device. also check for pause frames in the switches, if you don't have flow control on, test with it. it can help a lot in mix rate environments and busy links.
an example, we had a customer we installed 4 AF24s complaining they where only getting 250 meg from one of the links and there camera feeds kept cutting out all over the campus, but the capacity was showing 760s, when i got chance to look at it, a old UBNT device neggd 10mbps in the switch and was trying to carry camera traffic jamming up the switch, rebooting the busted M series radio fixed the bottle neck in the AF24.
from what you've posted about the cambium link, it looks clean and you should see more in the 80mbps range.
i have done a test from SM to the bts and i have the same result.
Now i have try another test from wireless speed test, on the AP there's 25/30mbit active from other users ,and the result is this :
16.161 Mbps
28.812 Mbps
so 16.16 + 25/30mbit is ever 45mbit...
What’s the code States and % DL used by those subs ?
what do you mean about % DL ?
| Total Frame Time Used | 97.8% |
this ???
anyway now i have do a ping from switch mikrotik to AP cambium and the result is ...
0 172.16.12.42 56 64 1ms
1 172.16.12.42 56 64 0ms
2 172.16.12.42 56 64 1ms
3 172.16.12.42 56 64 4ms
4 172.16.12.42 56 64 5ms
5 172.16.12.42 56 64 2ms
6 172.16.12.42 56 64 1ms
7 172.16.12.42 56 64 3ms
8 172.16.12.42 56 64 4ms
9 172.16.12.42 56 64 2ms
sent=10 received=10 packet-loss=0% min-rtt=0ms avg-rtt=2ms max-rtt=5ms
You think is something about CPU utilization on the AP?
97% of your downlink air time is used, thats your problem.
you've got a busy sub thats not at max modulation using up a share of the air time. you'll need to improve that users signal. if they are at say MCS 10, and the modem your testing with is at MCS15, both units are pulling equal amounts of network attention, your test CPE is going to see 1/2 if the printed throughput. 45 meg. while the MCS 10 user can only reach (20 meg i think) and while your pulling on the test radio, they fall to 10. its not that something is broken, your simply out of air time.
other brands try to balance the MBPS a second which would significantly impact your test radio, cambium however splits the airtime evenly among active users, stopping one bad link from killing the speed of an entire sector. you can make your test unit a high prioty sub and you should see the numbers your after, but be aware doing so will take a drastic amount of bandwidth away from your low priority subs if you don't have throttles in place. your best best is to just boost the code states of your busy, poor RSSI radios.
this is the screenshot of my users connected
anyway also in 40mhz i don't see no one difference ?
@Giuseppe wrote:this is the screenshot of my users connected
anyway also in 40mhz i don't see no one difference ?
you should improve those RSSIs before moving to 40 mhz, some of those are borderline dead..... getting 45 meg from this situtation...is well, really good.
the customers that are running in the high 70s and low 80s, if you can change them to force 200s or force 110s, or add reflectors, ect. you'll start to see your performance improve, and once you've got them to better states, say holding ALL MCS 13, switch to 40 mhz mode, those 13s will fall to 12 but you'll get a performance gain so long as noise allows. switching to 40mhz mode now will drop all of your 10s and 11s to MCS 9, or worse MCS 2 and 3. so youll end up loosing the benefits of mimo B as the radio falls to A to keep your links alive. you need better RSSI to hold higher code states on wider channels. if you want to experiment with your forecasted capacity, download the capacity planner spreed sheet in the downloads area, input your current setup and it will give you an expected result. frankly id expect the calculated result to be less than 45 meg. then change the weakest subs to a higher power (a force 200 will give you 11 points over the old integrated radio) and you'll see the expected performance improvement. the spread sheet assumes all CPES are active at the same time FYI.
improve those RSSIs and you'll be set.
https://support.cambiumnetworks.com/files/epmp/
the link planner and guide can be found here.
thanks for your help...
anyway i have done more test
i have improved more bandwidth with set MCS1 in the AP and framesize 5ms
and in udp test from my pppoe server to the SM i have 65mbps in download (80mbps loaded on the AP) and 37mbps in upload (50mbps on the ap).... same SM in tcp mode is 26mbps in download and 15 mbps in upload is that normal?
Giuseppe wrote:
thanks for your help...
anyway i have done more test
i have improved more bandwidth with set MCS1 in the AP and framesize 5ms
and in udp test from my pppoe server to the SM i have 65mbps in download (80mbps loaded on the AP) and 37mbps in upload (50mbps on the ap).... same SM in tcp mode is 26mbps in download and 15 mbps in upload is that normal?
is it normal that 5ms frame and control MCS1 is more perofrmance ? yes. the 2.5 ms frame drops your latency a decent amount, allows the epmp to be snyced with the old PMP 100 gear with some particular extra settings. the 2.5 ms frames looses about 10% of your bandwidth for the faster framming.
the management data being sent on MCS1 rather than 0 will use less airtime, freeing up more airtime for data traffic, another benifit (the default values are 5ms frame and MCS1 for management)
the UPD being faster than TCP is also normal, that large of a gap however can be a yes or no. some UDP test will shove the data down the pipe rather congestion or room for the traffic is there.... TCP has end to end control measures to ensure reliably delivery of packets, which is really good for the live network flow, but doesn't cram a pipe as well. that big of a difference, you'll still want to improve your downlink RSSIs on those subs, watch your retransmissions, and packet errors. takes steps to reduce and improve those and you'll see your TCP flow improve. TCP is a complicated protocol that does a good job making sure data gets to where its going.... UPD is a little more luck and a little less control.
good job on your improvements!!!
ok very very thank you for the help!!
Dear Concern .
If you show the client uses of a graph,than it is easy to us monitor client uses purpormance.
