Hello iota team, As a number of you been pushing towards pulling together the re-architected server stack, i’ve been working on the base build framework of similar— but more specifically for the ARM Cortex platform, using Raspberry Pi and OLinuXino hardware.
- We are coding the ingestion engine that polls physical points continuously. These are the digital + analog, I2C, SPI sensor and control points serviced by the card. - Will then ingest into in-memory key value table (UID | TIMESTAMP | VALUE) and be available for localized analytics and algorithms (defined and managed by the events/action engine) and then feed onto ZMQ engine to be fed upstream through the stack. Additional API interfaces that can be turned on will be RESTful and Websocket. QUESTION: - Currently testing using Spark, to keep analytics synergy with the approach being used in iota for servers. Currently functional and ingesting about 200 items per second on the Raspberry Pi 3. Is anyone else currently working on Spark for "micro server” topologies, and have any experiences to share? - Our use of Redis on the server build, for inmemory key values is working great. Exploring feasibility of either building simple inmemory table functionality ourselves vs perhaps Redis, if can get it light weight enough. Anyone aware of others further progressing Redis builds for micro servers? Before connecting the dots with code for iota lite, working to solidity other key tools in the tool chest beforehand. Thanks, -scott
