Ever since we have installed these routers 4+ years ago, we have had weird latency issues.
This is on a very lightly loaded network, each router not passing more than 100Mb on average.
We have 4 Mikrotik routers in total, all connected to different upstream providers with BGP.
Within our own network we are using BGP and OSPF.
The attached image shows our servers in our network for the same past 24 hours, you can see a range of patterns.
However, all of a sudden it will stop for a period of days, weeks or months and return to a flatline ping.
As an end user however, we never notice any impact, we only notice it with monitoring.
After extensive testing and liaising with Mikrotik, we drew a blank.
The most probable cause is when packets are sent to a CPU core that is processing BGP, which pretty much sits at 100% usage 24/7.
The monitoring service is provided by thinkbroadband.com (if you want to setup your own test, it’s free) and pings our servers every second.
Can anyone shed any light on this? Are you experiencing the same?
@telcouk, 3 of the charts are dated “06/05/2020”, all 3 are showing “Lateny (milliseconds)”, but all with different graphs.
How to interpret these charts?
Or are they of different devices, ie. monitoring targets?
Update: ok, got it, it’s about 4 devices.
You should try to bring the CPU load down, by selectively test-disabling some parts (protocols, services, firewall rules etc.) to pinpoint the cause.
Hi, thank you for taking time to respond. The only thing using the CPU is BGP (shown as routing in Profile), which unfortunately we can’t turn off.
The CPU load pretty much never goes above about 20% on any other single CPU core, with most sitting at 0%.
