Ever wondered how Netflix's Java apps shrug off 270,000 requests per second? A Netflix engineer, Hugo Marques, pulls back the curtain on the nitty-gritty, real-world struggles of optimizing high-scale, IO-bound data processing. He tackles messy issues like "Parallel Stream Greed" and those pesky Out of Memory errors you get from chaining async tasks.
The journey involves everything from classic Threads and CompletableFuture to tackling the "pinning problem" with Virtual Threads. Ultimately, the secret sauce for keeping things snappy and protecting dependencies comes down to smart use of Bounded Executors, rate limiters, and semaphores.
Watch on YouTube
Top comments (0)