Self - Scaling Registries

Architecture

Here is a high level architecture of how data flows in and out of the Self Scaling Registry platform.

Process for federated querying. User (A) logs into Query Server (B), and is authenticated by AAA service (C), receiving an Identity/Authorization token which Query Server passes to the Smart Aggregator (D) to discover subscribed studies associated with that user in the Overlay Service (E). User composes query using study ontology from Ontology Server (F). Query is compiled by Query Server and passes to Smart Aggregator, which verifies the user and query privileges, and rewrites the query with the Overlay Service-specified node group destination. Smart Aggregator signs and broadcasts the validated query to subscribed SPIN adapters in each destination i2b2 VM (G). Query is processed at each node and responses are passed back to Smart Aggregator which post-processes data from all responses. Aggregated response is then passed back to Query Server and user.

Process for data import. Initial data entry by site-based research coordinator(1) into web-based electronic data capture system(eCRF)(2). Sanitized eCRF data then flows into a data pipeline(3) that will deliver the data to the appropriate site specific ODM extract transform load (ETL) module(4). Each site has an i2b2 installation in its own virtual machine space(6B) hosted on a larger central server farm(6A), and connected together via SHRINE(5)