https://github.com/bert-r/streaming-fhir

Streaming FHIR implementation based on HAPI, Debezium and Kafka
https://github.com/bert-r/streaming-fhir

Last synced: about 4 hours ago
JSON representation

Streaming FHIR implementation based on HAPI, Debezium and Kafka

• Host: GitHub
• URL: https://github.com/bert-r/streaming-fhir
• Owner: Bert-R
• Created: 2023-06-07T19:07:47.000Z (over 1 year ago)
• Default Branch: master
• Last Pushed: 2024-12-17T14:02:19.000Z (about 1 month ago)
• Last Synced: 2024-12-17T15:18:53.480Z (about 1 month ago)
• Language: Java
• Size: 176 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Introduction

This repo implements a basic data platform for streaming FHIR resources. Every insert/update/delete on the HAPI FHIR

store is posted as a message on Kafka.

# Design

The chain starts from an interceptor in HAPI. That writes a message to the PostgreSQL write-ahead log (WAL) for every

insert/update/delete, containing the involved FHIR resource. This message is read by Debezium. A custom transformer in

Debezium converts the message to a simple FHIR resource, with a ``Struct`` key and posts it on Kafka. Resources in

the ``Patient`` compartment are posted on the ``patient`` topic, others on the ``nonpatient`` topic.

The interceptor in HAPI implements the outbox pattern: updates result in messages in the WAL, as part of the current

transaction. This results in exactly-once behavior: if the update succeeds, the message will be in the WAL. If it does

not succeed (transaction is not committed), there won't be message in the WAL.

Debezium implements at-least-once behavior: in case of failures, the same message might be issued more than once, but

you'll never miss a message.

# Running it

Run the script ``bin/setup`` to start everything. This enables the following interesting endpoints:

* [Adminer](https://www.adminer.org/), a simple database management tool: http://localhost:8081/ (login with System:

  PostgreSQL, server: postgres, username: admin, password: admin and database: hapi)

* [HAPI](https://hapifhir.io/), a FHIR store: http://localhost:8080/fhir/metadata

* [Kafdrop](https://github.com/obsidiandynamics/kafdrop), a web UI for viewing Kafka topics and browsing consumer

  groups: http://localhost:9000/

To post a new ``Patient`` resource, run ``bin/post-pat``, optionaly specifying the family name as parameter (defaults to

Duck). To update it run ``bin/put-pat``, optionaly specifying the resource ID and family name as parameter (family name

defaults to Trump). This will result in messages

in [the ``patient`` topic](http://localhost:9001/topic/patient/messages?partition=0&offset=0&count=100&keyFormat=DEFAULT&format=DEFAULT).

Note that HAPI skips identical updates, so subsequent calls to ``bin/put-pat`` with the same family name would not

result in Kafka messages.