3H/ Hybrid Personal Cloud – Applying devops open source tech to personal IoT
Hybrid Personal Cloud
Convener: Dave Sanford
Notes-taker(s): Dave Sanford
Discussion notes, key understandings, outstanding questions, observations, and, if appropriate to this discussion: action items, next steps:
Dave Sanford sketched on the white board and discussed the ideal enterprise model for 'hybrid cloud'; where systems are completely created by code/scripts (and data needed by that code/scripts) which is located on a different system than the one created by it. Another requirement of this model is for all of these system generating information (code, scripts, data) to be under version control, which enables roll-back etc. This creates a high level of control, autonomy and visibility for systems created and managed this way. In particular, they can be deployed in the cloud, on-premise, on new hardware on premise - with the push of a button. Horizontal scaling and avoidance of hardware and/or cloud service provider lock in becomes relatively easy. Disaster recovery comes for free. Dave indicated that his 'vision' of personal cloud would follow the same model - avoiding IoT hardware vendor lock-in, unknowable information sent by home devices to cloud vendors, etc.
Phil Windley talked about the fact that this model is hard, unless you have architectural assumptions that are shared with a larger community and indicated that this was one of the benefits of the picos (persistent compute objects). Phil also talked about the AWS 'Green Grass' model for IoT and how the corresponding System Development Kit (SDK) and corresponding 'device shadows' will allow API creating and loosely couple architecture for either accessing AWS services (e.g. Alexa application) or custom code behind a well defined API.
Dave talked about using the 'strangler' architectural pattern used to transform monolithic applications into microservices to use and interoperate with proprietary IoT device software components while maintaining the ability to replace or migrate away from them if appropriate.