I have a VPN in which (mainly two) clients communicate with each other. I realized that the throughput between two of the clients is extremely low ((2) -> (1) in the picture below) in one direction and later found out that this problem also occurs without a VPN (direct connection between two routers attached to the internet).
The following picture should help understanding my infrastructure:
where Ref is a reference to determine the maximum speed for each node (that where two speed test providers in my case).
(1) is a Raspberry Pi using latest Raspbian.
The throughput of the Raspberry Pi (1) is about 20 MBit/s in local network in either direction (wireless).
Throughput of (2) is also fine from any computer on the internet in any direction.
The combination described in the picture is the only combination known to me in which the bottleneck appears.
Latency is really stable between both nodes but I realized when pinging from (1) to (2) that icmp_seq
sometimes increases by more than one. Is that an indicator for packet loss?
PING 192.168.4.1 (192.168.4.1) 56(84) bytes of data.
64 bytes from 192.168.4.1: icmp_seq=2 ttl=128 time=58.10 ms
64 bytes from 192.168.4.1: icmp_seq=4 ttl=128 time=48.8 ms
64 bytes from 192.168.4.1: icmp_seq=6 ttl=128 time=71.5 ms
64 bytes from 192.168.4.1: icmp_seq=7 ttl=128 time=66.3 ms
64 bytes from 192.168.4.1: icmp_seq=9 ttl=128 time=58.4 ms
64 bytes from 192.168.4.1: icmp_seq=12 ttl=128 time=53.9 ms
Please let me know which information you need - for now, I have no idea where else to search for the problem.