shared-nothing-architecture

Shared-Nothing Pipeline: Three Independent Services in Sequence

In a shared-nothing architecture, each service owns its state and communicates only through explicit message passing -- no shared database, no shared memory. This workflow chains three services where each receives only the output of the previous service, then aggregates all results.

Workflow

requestId, data
       |
       v
+------------------+     +------------------+     +------------------+     +------------------+
| sn_service_a     | --> | sn_service_b     | --> | sn_service_c     | --> | sn_aggregate     |
+------------------+     +------------------+     +------------------+     +------------------+
  result: "a-processed"   result: "b-processed"   result: "c-processed"   aggregated from
  instanceId: "a-1"       instanceId: "b-1"       instanceId: "c-1"       all three
  (no shared state)       (uses a's output only)  (uses b's output only)

Workers

ServiceAWorker -- Processes independently with no shared state. Returns result: "a-processed", instanceId: "a-1".

ServiceBWorker -- Processes using only service A's output. Returns result: "b-processed", instanceId: "b-1".

ServiceCWorker -- Processes using only service B's output. Returns result: "c-processed", instanceId: "c-1".

AggregateWorker -- Combines all results into a final aggregated response.

Tests

8 unit tests cover each independent service and the aggregation.

Running

See ../../RUNNING.md for setup and execution instructions.