Does anyone measured the latency of ePMP with Force110 using as a Point to Point Link?
Hello, yesterday we install 3 PTP Links between 3 diferents sites, we noticed that the latency is really high in my screenshots you can see an increment of 14 ms for each hop. We really need this issue fixed thos links are 1 Mile.
- firsthop.png (8.16 KB)
- secondhop.png (7.74 KB)
Thanks for the quick feedback.
What is the DL/UL (75/25, 50/50, 30/70 or Flexible) ratio you are using?
What is the link quality and capacity on these links?
Link 1: DU is 50/50
Link 2: DU is 50/50
Link 3: DU 50/50 (This is the best link but the latency is higher here)
Using the 50/50 (or 75/25 or 30/70) ratios, the 14 ms latency per hop is expected. You may want to consider switching the links to Flexible mode to drop the latency to the 6-7ms range.
We will soon be coming out with a new software release (Beta expected to be out soon) that will drop the latency down to 2-3ms range for PTP links.
What is the timeline for this "soon" beta?
Hi Aero_Ramon, I'm not sure they explained this fully... You're probably clear on this, but maybe their answer will be be misleading for others reading it too, who might think that the 14ms latency is a 'problem' that needs a fix.
The situation is - if you setup the radios to use the 75% or 50% or whatever ratios, then that tells them that you want them to transmit and receive on a timed schedule. There are 'frames' where the Access Point is either transmitting, or it's listening, and the AP flips from TX to RX on a set schedule. That can be ideal for an Access Point with lots of Clients because it helps organize all the Stations (CPE's) and gives time-slots for everyone to talk without stepping on any other station. It also allows all your co-located Access Points to all transmit at the same time, and then to all go quiet and receive at the same time - and that can dramatically help with self-interference and with frequency-reuse. However, since it's on a schedule, a side-effect of that 'tick/tock' is that there will be a small amount of latency injected. The idea on a point-to-multipoint AP is that this will be predicable and stable latency.
On a point-to-point backhaul, you probably don't want to choose one of the ratios - you'd be better with the 'flexible' and non-scheduled setting. That allows the AP to transmit whenever it needs to, and it allows the Station side to transmit back whenever it needs to, and it allows more throughput in any one direction too. And, it'll dramatically lower the latency, because the AP and Station's aren't waiting to TX and RX on any schedule. It you pick 50%/50% schedule, then the AP is in 'listen' mode 50% of the time, even if there isn't data to be RX'd. And it's in TX mode 50% of the time, even if there isn't data to be TX'd. And if the AP is in TX mode, and if there is a packet ready to be RX'd, that data needs to wait till the end of the AP TX period, and wait till the AP switches into the RX period, before that data can be RX'd - and that 'tick/tock' timing is what injects that extra latency. It's not a problem, it's an option. It's designed for MultiPoint access point situations, where you want to have a scalable and organized way to have 50 Stations all have a organized schedule to talk back to the AP, and have all your AP's have an organized schedule to all talk at the same time so that they aren't deafening each other during the RX period.
We're shooting for this week. So keep checking back!