Lesson 01
Java Concurrency for Backend Systems Explained
Model tasks, threads, shared state, races, safety, and finite backend capacity.
Java Concurrency topic cluster
Java Concurrency & Async Programming
A production learning path for Java correctness, task execution, virtual threads, JVM and GC diagnostics, runtime capacity, and verified troubleshooting.
24 ordered lessons ยท Interactive concurrency lab
Connect Java correctness to runtime evidence and finite capacity.
Move from shared-state reasoning through execution, virtual threads, GC, memory, JIT, resource isolation, and operational proof.
Lesson 02
Java Memory Model, Visibility, and Happens-Before Explained
Separate visibility, ordering, atomicity, volatile handoff, and compound invariants.
Lesson 03
synchronized, Locks, and Atomic Variables in Java
Choose monitors, explicit locks, and CAS from the full invariant and cancellation boundary.
Lesson 04
ConcurrentHashMap Explained
Use documented concurrent map operations without assuming mutable values and workflows are atomic.
Lesson 05
ExecutorService, Thread Pools, and Work Queues in Java
Treat workers, queues, rejection, and shutdown as explicit admission and lifecycle policy.
Lesson 06
Java Interruption, Cancellation, and Deadlines Explained
Propagate cooperative stop signals without confusing caller timeout with effect rollback.
Lesson 07
CompletableFuture Composition, Errors, and Timeouts
Own stage execution, flattening, errors, timeouts, and underlying-work boundaries.
Lesson 08
Timeouts, Deadlines, and Cancellation Propagation Explained
Carry one end-to-end budget and preserve ambiguous durable outcomes.
Lesson 09
Java Virtual Threads for Backend Services
Scale blocking task representation while guarding CPU, connections, and downstream demand.
Lesson 10
Scoped Values and Structured Concurrency in Modern Java
Isolate Java 25 final Scoped Values from JDK 26 sixth-preview task structure.
Lesson 11
Bulkhead Pattern and Resource Isolation Explained
Partition finite failure and capacity domains so one workload cannot consume every resource.
Lesson 12
Backpressure, Bounded Queues, and Flow Control Explained
Bound admitted waiting and make queue age, expiry, and rejection visible.
Lesson 13
Load Shedding and Adaptive Concurrency Limits Explained
Reject excess work before overload destroys useful throughput.
Lesson 14
Production Java Concurrency Troubleshooting
Trace starvation, queues, locks, pinning, CPU, dependencies, and cancellation from user impact.
Lesson 15
Production JVM Performance Diagnostics Explained
Route user impact through CPU, GC, heap, native memory, threads, containers, dependencies, and durable outcomes.
Lesson 16
JVM Garbage Collection Logs and Pause Diagnosis Explained
Read allocation, live-set, pause, concurrent-cycle, promotion, and Full GC evidence without pause-only conclusions.
Lesson 17
G1 Garbage Collector Diagnosis and Tuning
Follow young, marking, mixed, evacuation, humongous-object, and adaptive-IHOP behavior before tuning.
Lesson 18
ZGC Low-Latency Garbage Collection Diagnosis and Tuning
Operate Java 25 generational ZGC through allocation headroom, stalls, CPU, heap, and latency-throughput evidence.
Lesson 19
Java Heap Dumps and Memory Leak Analysis
Move from live-set trends and histograms to protected heap-graph ownership and repeatable retention evidence.
Lesson 20
JVM Native Memory, Metaspace, and Container OOM Diagnosis
Reconcile heap, HotSpot native categories, RSS, cgroups, and OOM evidence with explicit NMT limits.
Lesson 21
JVM Thread Dumps, Safepoints, and CPU Saturation Diagnosis
Correlate platform and virtual-thread stacks, locks, safepoints, CPU, pinning, and dependency progress.
Lesson 22
JVM JIT, Code Cache, and Warmup Diagnosis
Diagnose tiered compilation, queues, deoptimization, code cache, warmup, JFR, and Java 25 AOT boundaries.
Lesson 23
Evidence-Based JVM Tuning and Capacity Verification
Turn one JVM hypothesis into a representative canary, rollback, capacity, user, and durable-state verification.
Lesson 24
Java Concurrency Budget Lab: Queues, Virtual Threads, and Cancellation
Run six deterministic capacity scenarios without treating the model as a benchmark or calculator.