Hi all, when using eptp for a 400c link is it a flexible mode or is it a fixed DL/UL percentage?
Thank you
The frame mode is flexible.
3 Likes
ePTP mode is ‘flexible frame’ and TDD mode is a fixed DL/UL %.
You can choose either ePTP or TDD modes for Point-to-Point links.
1 Like
Thank you both for your replies! Appreciated!! So like other epmp products if I were to do a link test while traffic is passing it should only show me what is available correct?
Eh, I’m not sure about that… I did a couple of tests on ePTP links and in both cases it appears to show the maximum link results including background traffic.
Okay, thank you! I figured but just was going by Cambium disclaimer in Maestro. So as per the flexible framing one could expect the radio to pull majority of DL if some UL space is available. As in if my aggregate of DL and UL is lets say 600mbps if 450 of DL is needed can this be achieved in flexible timing?
Anyone? If I get 600mbps aggerate, if I only have 16mbps of UL being used does the flexible allow more of the total be used for the DL?
Yes.
The way it work is this – in 5.0ms frame, there are 200 ‘time slices’ per second. In 2.5ms frame, there are 400 ‘time slices’ per second. Some types of radios have 4.0ms frames, or 2.0ms frames, and those would have 250 time slices or 500 time slices or whatever. SO, let’s assume you have a radio with 2.5 ms frames (which means 400 time slices per second) then you can then set either Fixed Frames of Flexible Frames to tell the AP what to do with it’s 1/400th of a second frames.
In a Fixed Frame, YOU choose the Ratio or ‘Duty Cycle’ and if that’s 75%/25% for example, then your radio with spend 75% of that’ 1/400th of a second in Download mode, and it’ll spend 25% of that 1/400th of a second in Upload mode. And it’ll switch from TX to RX exactly when you’ve told it to, regardless of what data is going over the link.
In Flexible Frame, it’ll still switch from both TX to RX in every 1/400th of a second frame, but it’ll fill as much of the frame with data, in whatever direction that data is going, to maximize the potential throughput.
So, if (just for easy math) if the overall aggregate capacity of a link is 800mbit, then in 75%/25% FIXED FRAM, you are telling the AP to spend 75% of it’s time downloading (so 600 Mbit max) and to spend 25% of it’s time uploading (so 200 Mbit max) even if there is more data than that — while the upside of Flexible Frame is YES, that it can do 750 Down & 50 Up, or 700 Down & 100 Up, or 780 Down & 20 Up, or 400 Down & 400 Up, or 150 Down & 650 Up – and it’ll dynamically fill the frames with as much data, going which every way it’s going.
The downside with FLEXIBLE is that you loose the benefits of synchronization – if you have multiple APs or SM’s close to each other, eith the RX/TX schedule being all ‘flexible’ you will have one AP RXing data right while some other nearby AP is still TXing data, and they will stomp on (interfer with) each other.
4 Likes
@ninedd, I really appreciate you taking the time to lay out an explanation on the Frame timing. In my base understanding of flexible, in which you have expanded, I understood that I could gain more DL if the UL was minimal which is the case for my 400c link. As I have been monitoring my link in busy hours I do not see the flexible happening. My link shows an aggregate of around 600mbps and in the past week during busy hours I am averaging 330mbps DL and 16mbps UL. I started getting messages from customers that they were experiencing slow speeds and as I investigated the only conclusion I came to is that my link is maxing out at 330mbps in flex mode. Thoughts? Here are some pics of stats…
The link is 5km long and I am running 5.4.2 on the link, We updated the link to 5.4.2 in December from 5.4.1rc15 and I am wondering if that is coincidence or not, we’ve hit over 300mbps in the past and did not hear from customers. But could be coincidence. Again I really, really appreciate you taking the time, Cheers!
What does your Monitor Performance say? You won’t get maximum speed unless all, or mostly all packets are going at maximum modulation (DS MCS11 or MCS10) which are the 1024-QAM modulations. Here’s one of mine – it’s not perfect, but it’s OKish at MCS10
But, if you look at yours and all the packets are going at MCS7 (for example) then that’s basically performing like an older generation radio and not really taking advantage of AX’s capabilities… not unless you get signals and SNR and Modulations up to MCS10 or MCS11
So, post a screen shot of your modulations from Monitor Performance.
1 Like
@ninedd ahh yes sorry I forgot to post that…
Master…
Slave…
I think they are not performing as well as they were previously… I must have some interference somewhere… but link test shows an aggregate of just under 600mbps… so theoretically the flex should help me get beyond 330mbps DL no? My rssi is -51/-52 and my SNR 44/45 so thats definitely not the issue.
Is there still an issue with the 400c doing this…
This is happening on the link in question.
@ninedd @Eric_Ozrelic Here is another link with a pair of 400c radios stats…
in a 40mhz channel does the above look to be accurate?
Still not sure if the flexible is working on these links.
Cheers!
Please update firmware to 5.6.0-RC26 and re-run your tests.
When you’re running the link test, is there around 100mbps of background traffic going over the link?
What ePTP mode are you using for this link? low latency, balanced, or high throughput?
Assuming most packets are being sent at 1024QAM, each 20MHz of spectrum should give you around 180-220mbps of tput depending on what ePTP mode you’re using.
1 Like
Thank you for your reply and I will implement the beta firmware and report back. We are currently using balanced.
Just for my 2c – in my testing, I didn’t see much difference unless the link was starting to get saturated. In other words, if a link is only 1/2 occupied, then it’s it’s usually already doing the entire throughput it’s asked to do, and it’s able to do it at a low latency.
What I found is if your link is getting close to full (approx. 400mbit @ 40Mhz width) then this can come into play and make a difference.
To me, this setting is really beautiful for tweaking how you want a more saturated link to behave… when the link get’s 90% full - you can decide if should sacrifice some latency for higher throughput, or should it sacrifice some throughput for lower latency.
Thank you for your reply! I am just not sure if the flex is working… I got one link that is good for 800mbps aggregate and one for 600mbps aggregate. What is best way to test it other than user traffic?
Use something that can generate 1gbps of traffic or higher, like 2 decently spec’d Mikrotik routers on each end of the link.
https://help.mikrotik.com/docs/display/ROS/Bandwidth+Test
Or use two high end laptops, one on each side of the link and use iperf.
@Eric_Ozrelic @ninedd thank you for your reply!
I ran a link test on the master 400c and it tested 400DL and 370UL. I have 2 mikrotik routers at each end of link and so I ran a bandwidth test from the slave side to the master side and when I set the slave side mikrotik to receive I saw 400mbps pass through the master. When I set it to send I saw 330mbps pass through the slave. So that is matching my Cambium link test… I am just not sure how to stress the ptp link to see the flex kick in for a single direction.
If I setup the bandwidth test to “receive” on the slave side I was under the impression the flex would kick in as I pulled from the slave and I would see it go above the 400mbps DL. I do apologize if I am not understanding correctly and open to correction so I can understand.
Really appreciate you taking the time!! Cheers!
I’m not sure what’s so difficult to understand here…
-
Do a test one way using 10 TCP streams. Watch the CPU usage on the Mikrotik’s to make sure they’re not maxing out. Make a note of the BW each way.
-
To test flexibility mode, do a bi-directional test, again using 10 TCP streams.
-
These 3 tests should show you the maximum traffic that you can transfer in each direction and in aggregate.
Okee doki…more characters to reply.