Some Key pointers & updates:
- Here in my problem statement, ive used aggregated Sensor value (min/max,avg , deviation) per day wise and get it stored in Neo4j and please note , not storing every second 'Telemetry' data to Neo4j since Neo4j is not meant for though it wont restrict you to store but it wud occupy more storage .
2. I did capture only the deviation (low/high ) and got the aggregated value for each type of sensor and connected to get the contextual insights
STEP 1: Build your Asset Hierarchy in Neo4j Graph Model - Instantiate through Cypher batch jobs
STEP 2 : Aggregate the Sensor values (outside Neo4j) for each device per day and only for some critical devices , took hrly aggregated sensor values and link with Day/Hr/Assets nodes but to avoid 'Super dense node' problem
ive managed with multiple relationship types by day wise/week wise/month wise
Take away:
1. In Simple Neo4j is best fit when there is many to many relationship and connecting different Entities/Things (Nodes)
and wants to traverse at 10..25..nth level from Top Nodes to Leaf nodes to get contextual Insights/patterns for better decision and predictions
2. Neo4j is not meant for storing just Telemetry data though you can actually store but other TS or Document db would be better for this purpose
3. Aggregated Metrics/Sensor values, we can take into Neo4j and definitely it would help us to get some known to hidden patterns when we connect different logical entities just not with only Telemetry data