The following diagram depicts the main components of the CluedIn application with a reference architecture on Microsoft Azure. You can also see a reference architecture directly on the Microsoft website.
CluedIn is made up of various functional layers, all encapsulated into their own Docker containers.
We have a combination of microservices running on .Net Core that handle various distinct functions, from handling the UI, to queuing and processing the large streams of data we need to ingest.
This handles how we connect securely to the cluster and handle permissions and grants to the different services the system is made up of.
- Neo4j - Manages complex relationships between CluedIn data objects (“Clues”).
- SqlServer - Manages our relational data storage. Alternatively a SAAS option (like SQL Azure) can be used.
- ElasticSearch - Indexes and searches the data for flexible querying. This can be scaled out as needed.
- RabbitMQ - The servicebus that handles queueing across the system.
- Redis - Used as the cache for the system.
Notice all the communication from the browser into the application comes through a set of ingress definitions (i.e. only a single public IP is required). The communication will all run over SSL in a production environment.
- For development and evaluation purposes using Docker
- For testing and production environments using Kubernetes