IoT devices are growing therefore more and more appliances starting from cars and machineries up to wearable such as watches are now smart and connected. This huge number of devices generates a huge number of messages which needs to be ingested, stored and processed. Classical ingestion mechanisms based on traditional servers are not suitable to sustain such a high volume and highly variable workload.
By leveraging on AWS services Reply has been able to implement a data ingestion platform fully based on AWS services, scalable and highly available. Reply solution is composed of modules:
Reply best practice in IoT Data Ingestion allows to ingest data via multiple channels and uses AWS Kinesis as collector for all data.
Once ingested data needs to be processed in order to identify events which might be combination of multiple events eventually generated by multiple devices:
example: when all lights are OFF and the thermostat is over 30°C turn the AC on or to conditionally forward messages to multiple consumer applications:
example: Send all temperature message to the thermal application, when for 10 times in a row the temperature is above 30°C send the message also to the alarm monitor application
Reply solution includes a Complex Event processor implemented on AWS Kinesis Data Analytics which is able to apply complex rules and handle new rules dynamically at scale being able to process incoming messages at very high frequency.
Moreover, in IoT latency might be very important for some use cases as we want to know that a specific alarm is on immediately, to satisfy this requirement, Reply solution has a configurable maximum latency which tunes AWS resources to comply with the specific requirement being able to achieve and end2end sub second latency event for high message volumes.