Skip to content

README.md

Latest commit

 

History

History
75 lines (48 loc) · 2.74 KB

README.md

File metadata and controls

75 lines (48 loc) · 2.74 KB

Netty Echo (Non-blocking I/O)

Demonstrates a client/server echo application built on Netty.

Netty uses an event-driven non-blocking model where channels are handled by event loops instead of one dedicated blocking thread per connection.

Core concepts

Channel

A Channel represents an open connection that can perform asynchronous read/write operations.

Future

Netty extends Java future-style control flow with ChannelFuture, so operations can be observed or chained without blocking the current thread by default.

Events and handlers

Netty pipelines process inbound/outbound events through handlers. Typical inbound events include:

  • channel active/inactive
  • read events
  • exceptions
  • user events

See implementation details in netty-server/src/main/java/my/javacraft/echo/netty/server/NettyServerHandler.java.

Encoders and decoders

Protocol payloads are transformed by pipeline codec handlers. This project uses line-delimited frames with string decoder/encoder handlers on both client and server sides.

Modules

  • netty-client: Netty client implementation and client-side tests.
  • netty-server: Netty server lifecycle and protocol handling.
  • netty-testing: Cucumber functional/load/benchmark scenarios.

Cucumber Performance Benchmark

Performance scenarios are implemented in netty-testing/src/test/resources/features/netty-benchmark.feature.

  1. NettyServer + NettyClient benchmark:
mvn -pl echo/netty/netty-testing -am test -Dcucumber.filter.tags='@Performance'
  1. Final summary (reads persisted benchmark summary from target/performance-results):
mvn -pl echo/netty/netty-testing -am test -Dcucumber.filter.tags='@PerformanceSummary'

The benchmark flow includes warmups (not measured), 3 measured runs, and prints average/median metrics.

Example Final Result

[PERF][FINAL] average execution time for netty benchmark case:
[PERF][FINAL] 1) NETTY_SERVER+NETTY_CLIENT average: 0.206 s, throughput=48483.37 msg/s, avg=0.0206 ms/msg, delta=0.000 s (+0.00% vs fastest)
[PERF][FINAL] total measured execution time: 0.619 s (single test)
[PERF][FINAL] total benchmark scenario time (warmups + measured): 1.478 s (single test)

Why This Is Expected

  • Netty handles many sockets through event-loop scheduling instead of one blocking thread per client.
  • Line-based codec and pipeline handlers keep request/response processing lightweight for this protocol.
  • Warmups reduce first-run noise (JIT/classloading), so measured runs are more stable.

Important Notes

  • Results are environment-specific (CPU, OS scheduler, JVM version, background load).
  • Compare benchmark numbers only on the same machine and runtime configuration.