Advanced Mule 4 Development (Course #advanced-mule4)

Programme every line is something you do

8 modules · 24 topics · in course-flow order

1. Module 1

   DataWeave at Memory Scale

   Objective. Refactor the settlement mapper into streaming DataWeave 2.x functions and modules that keep heap under a target 512 MB.

   • Write reusable DataWeave 2.x functions and modules, exercised standalone before wiring into flows
   • Stream oversized payloads instead of buffering them, proving flat memory on a 50 MB ISO 20022 file
   • Cut heap pressure: compare streamed vs buffered mappings in VisualVM (target: heap < 512 MB)
2. Module 2

   Batch, Async & Cache Levers

   Objective. Construct a watermarked Batch job settling 1 M transactions overnight, plus VM/Anypoint MQ async lanes and cache scopes.

   • Assemble Batch jobs: scope, steps, aggregate-report, on-complete, and watermarks persisted in Object Store v2
   • Decouple slow systems: VM queues for in-app async, Anypoint MQ for durable cross-worker delivery
   • Size cache scopes (TTL, key expression) around reference data (target: 60% fewer downstream lookups)
3. Module 3

   Custom Connectors & Java

   Objective. Arbitrate when a custom SDK connector pays off, then wrap Meridian's HSM signing library into a working connector.

   • Arbitrate build vs reuse: custom SDK connector vs HTTP wrapper vs scripting module, maintenance cost included
   • Scaffold a Mule SDK connector on Java 17: @Operations, @Configuration, @Connection management
   • Invoke Java from flows: Java module, static methods, POJO payloads, classloader pitfalls
4. Module 4

   Profiling to Throughput

   Objective. Diagnose the payment bottleneck with JFR and VisualVM, then retune workers toward 2,000 msg/s at a target CPU under 70%.

   • Profile with Java Flight Recorder and VisualVM: blocked pools, GC pauses, logging overhead
   • Keep production paths lean: streaming readers, object-store eviction policy, heap headroom targets
   • Tune throughput: max-concurrency, worker/vCore sizing on CloudHub and RTF, queue depth math
5. Module 5

   Resilience by Design

   Objective. Engineer a retry topology — until-successful exponential backoff + jitter, circuit breaker, DLQ — that rides out core-banking outages.

   • Model error types and propagation: CUSTOM errors, rethrow semantics, global vs flow-scoped handlers
   • Wrap flaky COBOL calls in until-successful with exponential backoff + jitter and reconnection policies
   • Divert poison messages to an Anypoint MQ DLQ, replaying them safely via the idempotency-key store
6. Module 6

   MUnit Beyond Happy Paths

   Objective. Deliver a MUnit suite mocking every downstream system, exercising error branches, gating builds at a target 85% coverage.

   • Mock and spy connectors: mock-when with DataWeave-generated responses, spy-verify on outbound calls
   • Set meaningful coverage gates (target: 85%) and wire reports into the Maven build as a hard gate
   • Simulate timeouts and HTTP 500s so backoff, circuit-breaker and DLQ paths prove themselves in CI
7. Module 7

   Real-Time Push at Scale

   Objective. Design a WebSocket push channel broadcasting fraud alerts to 5 k concurrent partner clients across workers.

   • Operate the WebSocket connector: server vs client sockets, session lifecycle, heartbeat and reconnect
   • Fan out pub/sub: bridge Anypoint MQ fraud-alert subscriptions to every connected WebSocket session
   • Plan scale-out: sticky sessions across 2+ CloudHub/RTF workers and the cost of stateful connections
8. Module 8

   Coherence Under Failure

   Objective. Guarantee exactly-once money movement between Meridian's ledger and payment rails even when middleware dies mid-flight.

   • Scope local vs XA transactions across Database and JMS endpoints; arbitrate when two-phase commit earns its latency
   • Tune Anypoint MQ ack modes so redelivered initiations hit the idempotency-key store instead of double-charging a customer
   • Prove recovery: kill a worker mid-batch, replay from the DLQ, reconcile zero duplicates and zero losses in the ledger

References & Official Documentation every claim above has a source

Module 1

• Streaming in Mule AppsMuleSoft Docs

Module 2

• Batch ProcessingMuleSoft Docs

Module 3

• Mule SDKs OverviewMuleSoft Docs

Module 4

• Back-Pressure and MaxConcurrencyMuleSoft Docs

Module 5

• Until Successful ScopeMuleSoft Docs

Module 6

• MUnit CoverageMuleSoft Docs

Module 7

• WebSockets ConnectorMuleSoft Docs

Module 8

• Transaction ManagementMuleSoft Docs

More in Developer