🍃☕ Forget the 10X DEVELOPER. Build a 10X JAVA APP
🍃☕ Forget the 10X DEVELOPER. Build a 10X JAVA APP
Same code. Same blocking call. One property. ~10× less wall time.
🔸 TLDR
▪️ 1 Tomcat worker thread
▪️ 🔟 concurrent requests
▪️ Each request blocks for 1 second
▪️ With Platform threads: 10.0863 seconds
▪️ With Virtual threads: 1.0652 seconds 🚀
That is roughly 9.5× faster in this benchmark.
And the application code does not need to become reactive.
🔸 THE BLOCKING ENDPOINT
@GetMapping("/benchmark") public BenchmarkResult benchmark() throws InterruptedException { Thread.sleep(1000); // Simulates blocking I/O Thread currentThread = Thread.currentThread(); return new BenchmarkResult( "Blocking work complete", 1000, currentThread.getName(), currentThread.isVirtual()); }
Thread.sleep() represents the kind of waiting we often get from:
▪️ Database calls
▪️ HTTP APIs
▪️ File I/O
▪️ Messaging
▪️ Other blocking operations
With a traditional platform thread, that thread remains unavailable while the request waits.
🔸 FORCE TOMCAT TO ONE WORKER
server.tomcat.threads.max=1
server.tomcat.threads.min-spare=1
Now send 🔟 requests concurrently:
hey -n 10 -c 10 http://localhost:8080/benchmark
With one traditional worker processing ten 1-second blocking requests:
Total: 10.0863 secs
They are effectively processed one after another. 🐌
🔸 ENABLE VIRTUAL THREADS
With Java 21+ and a compatible Spring Boot version:
spring.threads.virtual.enabled=true
Run the exact same benchmark again:
Total: 1.0652 secs
🔥 Nearly all 🔟 blocking requests can now spend their waiting time concurrently.
Instead of tying expensive OS-backed platform threads to blocking operations, Java can park lightweight virtual threads and reuse the underlying carrier threads.
🔸 WHY THIS MATTERS
Virtual threads let you keep the familiar imperative Java programming model:
result = repository.findById(id);
response = apiClient.call(result);
return response;
without requiring you to rewrite everything using reactive chains just to achieve high concurrency.
But there is an important nuance 👇
Virtual threads do not make CPU work 10× faster.
They shine when your application spends significant time waiting on blocking I/O.
🔸 TAKEAWAYS
▪️ Virtual threads dramatically improve scalability for blocking workloads.
▪️ In this deliberately constrained benchmark, ~10 seconds became ~1 second.
▪️ Your synchronous Spring MVC code can stay synchronous.
▪️ They reduce the cost of having many concurrent waiting tasks.
▪️ They are not a magic CPU-performance switch.
▪️ For many traditional Spring Boot applications, virtual threads make thread-per-request attractive again. ☕🚀
#Java #SpringBoot #Spring #VirtualThreads
Go further with Java certification:
Java👇
Spring👇
SpringBook👇
JavaFullstackBook👇