Compare Flow with Disruptor

[dansiviter]

Compare if the 90KB+ dependency of com.lmax:disruptor is worth it over Flow.

[dansiviter]

Performance isn't what I expected out of the box, Disruptor is roughly the same as Flow:

Benchmark                                                              (handlerName)   Mode  Cnt      Score       Error  Units
ConsoleHandlerBenchmark.handler                                 j.u.l.ConsoleHandler  thrpt   25  25365.926 ±  2550.259  ops/s
ConsoleHandlerBenchmark.handler:Used memory heap                j.u.l.ConsoleHandler  thrpt   25  44632.638 ±  8489.078     KB
ConsoleHandlerBenchmark.handler                            u.d.l.AsyncConsoleHandler  thrpt   25  74252.157 ±  5302.526  ops/s
ConsoleHandlerBenchmark.handler:Used memory heap           u.d.l.AsyncConsoleHandler  thrpt   25  34605.517 ± 14178.321     KB
ConsoleHandlerBenchmark.handler                   u.d.l.DisruptorAsyncConsoleHandler  thrpt   25  73753.890 ±  6732.131  ops/s
ConsoleHandlerBenchmark.handler:Used memory heap  u.d.l.DisruptorAsyncConsoleHandler  thrpt   25  34410.716 ± 14086.516     KB

Therefore the library size penalty isn't worth it. Disruptor may be configured to out perform but that goes against the ethos of lean and light.