Stream data in real time from aws to other devices.

The closest to real time would be to have the user subscribe to the topic and process the received messages. This will induce the minimum delay. The delay will be measured in 100s of milliseconds.

Note that 100 TPS is the maximum limit of messages being sent, and only for QoS 0. If the inbound message flow can exceed that (say from 2 or more devices), using another AWS service such as Kinesis Data Streams or Timestream+Grafana may be a better solution, but will increase the latency by 100s to 1,000s of milliseconds.