I’d like to minimize network traffic between embedded gateways and a cloud instance of InfluxDB for the purposes of running on a cellular network. The original data is sourced from 1m queries over local SCADA type networks. It could be logged onto an embedded InfluxDB container at the 1m intervals and then CQ’d into 15m tables before syncing offsite via a TICK script.
David Simmons discusses a similar concept in a blog post but neglects to mention how connectivity issues might be handled. He uses Kapacitor to both downsample and forward the stream offsite through a script. What happens when the network fails to connect and data starts to back up, will Kapacitor be capable of managing the data backlog? Does anybody have suggestions how best to handle network connectivity issues?
I am making the assumption that using the line protocol is more efficient than MQTT in order to transport data offsite; however, if someone thinks otherwise I’d be interested in their experience.