This use case demonstrates how to implement the SAGA Pattern to achieve distributed transactions across multiple microservices, assuming the implementation utilizes a Database per Service approach. The solution employs the Transactional Outbox Pattern in conjunction with Change Data Capture (CDC) and Debezium on top of Kafka Connect, and, Apache Kafka serves as the messaging backbone for the overall architecture.
Use case: As a guest, I want to make a hotel reservation
There are 3 microservices involved:
Reservation Service- initiator and orchestrator of the SagaHotel Service- ensure the hotel's room requested is available and book it or rejects it if unavailablePayment Service- executes the payment associated to an incoming reservation
And we have the following service configuration:
Use the SAGA Orchestration Pattern to ensure data consistency between services.
The reservation process is as follows:
The SAGA framework state machine happy path:
A detailed component workflow:
A happy path for making a reservation is illustrated in the following sequence diagram:
An unhappy path of making a reservation, when the payment is rejected, and the compensation step is involved you can see in the following sequence diagram:
Start the docker compose (docker-compose.yaml)
/src % docker-compose up --buildDuring the build process there is a kafka-setup service which creates kafka topics and submit the debezium connectors
Make a reservation
% http POST http://localhost:8080/api/v1/reservations < e2e/room-reservation-placement.json
HTTP/1.1 202
Location: http://localhost:8080/api/v1/reservations/6656d80e-edf0-42ab-b114-ced902bafa36
Retry-After: 0.5Examine the emitted event for room-booking.inbox.events in Apache Kafka:
% docker run --tty --rm \
--network go-saga-orchestration-network \
quay.io/debezium/tooling:1.2 \
kafkacat -b kafka:9092 -C -o beginning -q \
-f "{\"key\":%k, \"headers\":\"%h\"}\n%s\n" \
-t room-booking.inbox.events
{"key":057ceada-02b3-4a65-beb3-3de54d6e29f3, "headers":"id=aed1882e-c6d2-402c-985f-e40d19d4df43,eventType=REQUEST"}
{"type":"REQUEST","roomId":1,"endDate":"2023-12-17","guestId":10000001,"hotelId":1,"startDate":"2023-12-16","paymentDue":1702632793441,"creditCardNo":"************7999","reservationId":"6656d80e-edf0-42ab-b114-ced902bafa36"}and, the emitted event for payment.inbox.events in Apache Kafka:
% docker run --tty --rm \
--network go-saga-orchestration-network \
quay.io/debezium/tooling:1.2 \
kafkacat -b kafka:9092 -C -o beginning -q \
-f "{\"key\":%k, \"headers\":\"%h\"}\n%s\n" \
-t payment.inbox.events
{"key":057ceada-02b3-4a65-beb3-3de54d6e29f3, "headers":"id=e3280316-16b8-4114-98ee-9443d523d2ea,eventType=REQUEST"}
{"type":"REQUEST","roomId":1,"endDate":"2023-12-17","guestId":10000001,"hotelId":1,"startDate":"2023-12-16","paymentDue":1702632793441,"creditCardNo":"************7999","reservationId":"6656d80e-edf0-42ab-b114-ced902bafa36"}Query the saga state in the reservation service's database:
% docker run --tty --rm -i \
--network go-saga-orchestration-network \
quay.io/debezium/tooling:1.2 \
bash -c 'pgcli postgresql://reservationuser:secret@reservation-db:5432/reservationdb'
SELECT * FROM public.sagastate;
+--------------------------------------+-----------+------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+----------------+-------------------------------------------------------+---------------+
| id | version | type | payload | current_step | step_status | saga_status |
|--------------------------------------+-----------+------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+----------------+-------------------------------------------------------+---------------|
| 057ceada-02b3-4a65-beb3-3de54d6e29f3 | 3 | room-reservation | {"type": "REQUEST", "roomId": 1, "endDate": "2023-12-17", "guestId": 10000001, "hotelId": 1, "startDate": "2023-12-16", "paymentDue": 1702632793441, "creditCardNo": "************7999", "reservationId": "6656d80e-edf0-42ab-b114-ced902bafa36"} | | {"payment": "SUCCEEDED", "room-booking": "SUCCEEDED"} | COMPLETED |
+--------------------------------------+-----------+------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+----------------+-------------------------------------------------------+---------------+And, also, query the location URL:
% http GET http://localhost:8080/api/v1/reservations/6656d80e-edf0-42ab-b114-ced902bafa36
HTTP/1.1 200
Content-Type: application/json
{
"guestId": 10000001,
"hotelId": 1,
"reservationId": "6656d80e-edf0-42ab-b114-ced902bafa36",
"roomId": 1,
"status": "SUCCEED"
}
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent