"Investigators scoured more than 3,000 channels of video and telemetry data covering a very brief timeline of events – there were just 93 milliseconds from the first sign of anomalous data to the loss of the second stage, followed by loss of the vehicle."
I haven't seen anything about latency--are you sure that's a problem in the telemetry stack?
> SpaceX’s new implementation (for Falcon 9 “Full Thrust” flights) of non-
deterministic network packets in their flight telemetry increases latency, directly resulting in
substantial portions of the anomaly data being lost due to network buffering in the Stage 2 flight computer.
However it looks that finding might've actually been corrected in the months before Amos-6, so moot point!
SpaceX also had an architecture that added a lot of latency to their telemetry transmission (IIRC basically Ethernet bufferbloat)