Diagnosing unexpected websocket data usage on Hologram

I’ve run across an issue that may or may not be Hologram related. I have several deployed devices communicating with a server via JSON over Websocket. Some of these devices use locally available wifi, and others use Hologram SIMs. I’m seeing a drastic difference in the data use of these devices depending on which is connected, as well as intermittent spikes in the data usage of the SIM connected devices. I’ve attempted to correlate the data use reported on Hologram with that sent and received by my servers, but the numbers I have imply a greater than 90% overhead, which may be atypical for websockets. Even after making extremely conservative assumptions about sources of overhead (50 KB/hour ping pongs, frequent reconnections, etc), there’s still a third of my data use I absolutely can’t explain. During the spikes this unexplained use is well above 90%. For WiFi these numbers are incomplete but better - in particular we see the websocket disconnections, which are frequent with Hologram, are much less common.
This leads to two questions: What, if anything, about using cellular data would cause elevated data usage on a websocket, and if the device was doing something unexpected on the cellular network (phoning home, pulling firmware, etc) how would I detect that? I can’t modify the devices in question and the device logs are useless, so any observation needs to be at the network level. Is there a simple way to determine the destination of our outgoing traffic through Hologram?

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.