Deployment

There are two options to deploy Knowlg:

Option 1: Independently deploy Knowlg building block

The necessary documentation for deploying Knowlg independently is currently in the works. You can reach out to us in the community forums and we will assist you to complete the deployment.

https://github.com/Sunbird-knowlg/Community/discussions

Option 2: Deploying Knowlg as part of Sunbird-ED

Please refer to the Sunbird-Ed deployment for more details.

Sunbird ED

SB-Knowlg is packaged with some micro-services as shown in the above architecture.

• Content-service

• Search-service(also called asset-search-service)

• Taxonomy-service

• DIAL-service

The below list of DB's are used in the above micro-services

• Neo4J: This is used as a primary data store of the content-service & Taxonomy-service. The data of each asset is stored as a graph node. The data from this Neo4j DB will sync to Elasticsearch through flinkJob(search-Indxer) for ease of discovery.

• Cassandra: This DB is used by content-service, Taxonomy-service & DIAL service. Content-service: The asset data is stored in Neo4J. B the external/additional data of the asset will be stored in Cassandra because of large size data/information like Body, Collection hierarchy etc. Taxonomy-service: Object category definations are stored here. DIAL-service: We are using Cassadra as a primary store of DIAL data/information. This data will be synced to Elasticsearch through flinkJob(search-Indxer)

• Redis: To improve the performance of API instead of hitting the DB for every Read call, we are using the Redis as temp cache.

• Elasticsearch: This is the primary DB of the search-service(asset-search-service). There is not directly feed/post of the data to the elastic-search DB. The data will be synced from Neo4J to elasticsearch through Flink job(Search-indexer) only.