'Ping overlay' is used to validate the VXLAN tunnels, but the rtt value in its output may not be accurate to determine delay/latency.
The following description is from Juniper's document on ping overlay :
When this command is used, we see more than 1 msec round trip delay time (rtt) during ping overlay via VXLAN. The rtt value varies quite a bit when the command is issued several times.
user@router> ping overlay tunnel-src 32.68.230.85 tunnel-dst 32.68.230.83 vni 110091 ping-overlay protocol vxlan vni 110091 tunnel src ip 32.68.230.85 tunnel dst ip 32.68.230.83 mac address 00:00:00:00:00:00 count 5 ttl 255 WARNING: following hash-parameters are missing - hash computation may not succeed end-host smac end-host dmac end-host src ip end-host dst ip end-host input-ifd-idx end-host protocol end-host l4-src-port end-host l4-dst-port Request for seq 1, to 32.68.230.83, at Mar 07 2019 11:03:44.234 EST Response for seq 1, from 32.68.230.83, at Mar 07 2019 11:03:44.341 EST, rtt 109 msecs Overlay-segment present at RVTEP 32.68.230.83 End-System Not Present Request for seq 2, to 32.68.230.83, at Mar 07 2019 11:03:45.235 EST Response for seq 2, from 32.68.230.83, at Mar 07 2019 11:03:45.241 EST, rtt 17 msecs Overlay-segment present at RVTEP 32.68.230.83 End-System Not Present Request for seq 3, to 32.68.230.83, at Mar 07 2019 11:03:46.236 EST Response for seq 3, from 32.68.230.83, at Mar 07 2019 11:03:46.276 EST, rtt 134 msecs Overlay-segment present at RVTEP 32.68.230.83 End-System Not Present Request for seq 4, to 32.68.230.83, at Mar 07 2019 11:03:47.237 EST Response for seq 4, from 32.68.230.83, at Mar 07 2019 11:03:47.252 EST, rtt 18 msecs Overlay-segment present at RVTEP 32.68.230.83 End-System Not Present Request for seq 5, to 32.68.230.83, at Mar 07 2019 11:03:48.237 EST Response for seq 5, from 32.68.230.83, at Mar 07 2019 11:03:48.263 EST, rtt 74 msecs Overlay-segment present at RVTEP 32.68.230.83 End-System Not Present
This is expected behavior. Ping overlay is a good way to check for link availability. However, it is not the most accurate way to test latency or delay. For example, a more accurate way to test for latency is to simulate data plane traffic using a traffic generator for the transit traffic.
There is some reasoning behind the RTT value for ping and why it may not be accurate. Juniper Networks platform architecture separates the control plane from the data plane. There are various rate-limiting and prioritization functions within the packet forwarding engine (PFE) and the routing engine. When a ping is sent from the routing platform to the end device, it uses the control plane on the local router in order to generate the ICMP request packet. The control plane is used again when the ICMP reply packet is received from the end device.
ICMP messages are considered low priority within JUNOS, so the routing platform will respond to and process other higher priority messages, such as routing updates, before processing the ICMP messages. The microkernel may introduce tens of milliseconds of processing delay to ICMP message handling. The delay is not uniform, meaning that some ICMP messages might be delayed while others may not be delayed.