Changes in the backend: Much better now!

Yesterday, 18.11.2019, 11.00h, TTN changed something in the MQTT-configuration. This has a dramatic influence. There have been several post (1, 2, 3) about data losses last time.

We have been logging packet losses and travel time on the backend (Time difference between gateway reception and mqtt delivery) during the last weeks, and this are the results (Last two days):


This is the same plot for 20 days:

The green line gives the average travel time on the backend, which was between 3 and 10s.
For the loss rates we analyzed the packet counter and the normalized time between two packets.

As you can see, losses in TTN stopped completely yesterday. The travel time now is about 300 ms, which is merely the delay caused on the internet connection between gateway and TTN.


Do you know when the problems did start? Do you have data for that?

No. My first post about packet loss was Sep 4, but there it was not clear if it´s an UDP-issue. It took some time to digg a bit deeper. There is a post about TTIG packet loss dated Jul 18, maybe caused by this too.