1 of 1 people found this helpful
I would agree that is a bit much. Other customers production experience with non-batching and non-blocking sends are on the order of 10 microseconds or less in the send call.
Much of the behavior of UM is directly related to the resources available to library and what other applications are contending for those resources.
I would note that to measure latency, we recommend using lbmping and measure round trip time. The time spent in lbm_src_send() does not reflect the actual latency between the send call and the time that the packet hits the wire.
I would suggest that you raise a support request. We will need some details on exactly how you are measuring the time, your UM configuration, and some details on the hardware on which you are running the test.