Strange ping behavior, please help

I can not find an explanation for this. I would appreciate any guidance.

I have the typical scheme:
Client PC → SXT —> Omnitik → x86 RouterOS → Cloud

I make an endless ping to an IP test (DNS IP) from:
From PC client: Always 20 to 40ms
From SXT: always between 20 and 40 ms
From Omnitik: always between 20 and 30 ms
RouterOS x86: normally 20 to 30ms, BUT sometimes between 500-700, even timeout

I find that some PC clients/XBOX with a specific online games, and with ping between 20 and 30ms to DNS IP, can not play correctly because their ping time recorded in the game are exactly the same that records x86 RouterOS (500-700)
I can not find a pattern when I have 500-700. It is usually in the evenings, but even with consumption rates bandwidth and low cpu.

How is it possible that while the ping from the client or from the SXT is 20ms, from x86 Router is 700ms at same time?

Any idea that can happen?
(I’ve changed the x86 hardware)

Please, any advice? :frowning:
I changed all … switches and x86 RouterOS.
How is it possible this behavior?

x86 have only one ethernet interface (and multiple vlans), It can influence?
(No TX/RX drops, No TX/RX errors)
CPU Load low, max at 40%, 15% normally