【IoTDB Meetup】Pragmatic minds

1 . pragmatic minds GmbH Digital Twin with IoTDB Julian Feinauer & Simon Mayer 

2 .Some Background • Our ”Things” are (big) machines • Our customers are machine manufacturers or machine operators • Our customers are interested in lots of data • We integrate IoTDB into different tools and systems • Visualization / Dashboards • Analytics • Business Logic • IoTDB is used at the edge and in the cloud 

3 .Architecture Overview Manufacturer Company 1 Company 2 

4 .Some Numbers • Around 500 metrics • Sampling times between 10ms and 1s • Users can add new metrics anytime • Metrics are fetched using Apache PLC4X 

5 .Ditto based Twin 

6 .Eclipse Ditto 

7 .Eclipse Ditto based Architecture 

8 .Security and Access Control 

9 .How does it work? 

10 .Ditto based Authorizer 

11 .IoTDB meets Django 

12 .• Django based backend • We integrate multiple tools like – NodeRed – NextCloud – Grafana – Prometheus – ... • Access is managed in Django Application • Many metadata to each machine and metric 

13 .Architecture Overview Manufacturer Company 1 Company 2 

14 .Data Model TelemetryValue id AutoField • We store many metadata to each machine ForeignKey (id) app_name CharField machine name CharField • We also store metadata to each machine (telemetry) TelemetryValue • We also store Information about access Machine id AutoField control owner ForeignKey (id) Employee cockpit_id CharField user OneToOneField (id) description CharField company ForeignKey (id) lat FloatField email CharField location CharField is_company_admin BooleanField lon FloatField name CharField machine_id CharField name CharField type CharField owner (machines) company (people) user (employee) Company id AutoField User name CharField 

15 .Application Flow We implement the same REST API as the IoTDB Grafana Bridge 

16 .

17 .Short Summary 

18 .Summary • Two different approaches to model digital twins • Ditto based approach – Intersect IoTDBs Permission logic / API to perform checks against a 3rd party system – Good • Extremely flexible • Transparent for the user (can use plain IoTDB APIs like Session or JDBC) – Bad • Complex and error prone • Needs custom code linked into IoTDB 

19 .Summary • Two different approaches to model digital twins • Django based approach – Create one (or more) APIs in front of IoTDB – Good • Easy and extensible • Can also add metadata layer on top of IoTDB – Bad • Need to maintain a custom API • Probably multiple APIs needed for multiple systems