Although Java might not be the trendiest programming language presently, it’s nonetheless a strong alternative for a lot of, together with Netflix. There have been developments for builders to make AI-infused Java purposes as nicely.
Java isn’t any extra perceived as a legacy programming language. It has developed through the years. And Netflix has pushed its stack ahead—previous JDK 8, past reactive programming, and new tech.
Paul Bakker, a workers software program engineer at Netflix, shared insights about how they’re utilizing Java in his discuss on ‘How Netflix Makes use of Java’ at JavaOne 2025. “Lots of people apparently are usually not so pleased with Java. Some folks by no means need to watch Netflix once more as a result of it’s tainted by Java,” he mentioned.
Two Sides of Netflix, All Powered by Java
Netflix has two sides of a tech stack powering its operations. The primary powers its international streaming service, a heavy-traffic system unfold throughout areas the place one request can fan out into dozens. For this, Bakker mentions that their strategy is that, when one thing fails, a retry may resolve it.
The second half entails Netflix Studios, which is enterprise-focused. These handle movie manufacturing workflows — schedules, gear, folks, and logistics. Right here, information integrity is paramount. As Bakker put it, “If somebody saves information a couple of film being deliberate, that information completely must persist.”
Bakker harassed that Netflix’s structure for each the necessities stays primarily the identical, all powered by Java.
Upgrading The Backend Java Stack
Java Improvement Equipment is the software program improvement atmosphere that allows builders to create and run Java purposes. The migration to fashionable Java wasn’t glamorous, nevertheless it was vital. The corporate patched the legacy libraries, received the providers to make use of JDK 17 and above to step up a notch.
For anybody in the identical boat seeking to improve issues, Bakker steered, “Simply repair it. It would look arduous, however it’s really not.” He highlighted that they did nothing particular right here, however solely upgraded. With that alone, Netflix skilled vital efficiency enhancements.
“We received about 20% much less CPU time spent on rubbish assortment,” Bakker famous. He emphasised that it was a giant win for Netflix to attain a 20% efficiency increase on their machines with none tuning, which is commonly a difficult course of.
With JDK 21 and the introduction of generational ZGC, issues improved much more: “We went from greater than a second pause instances to zero.” He highlighted, “Considerably fewer errors on our IPC calls.”
Bakker was fairly blunt about REST and mentioned, “I don’t assume it’s best to use REST in any respect.” He put forth the argument that GraphQL has extra advantages comparatively for Netflix.
Bakker praised one other new expertise with the Java tech stack improve, digital threads. He defined that they allow processes to run in parallel, giving them improved efficiency by default.
Throughout experimentation, they encountered just a few points with it. Nevertheless, he talked about {that a} repair arrived in JDK 24, and Bakker’s staff is assured that it’s going to get higher.
Counting on Spring Boot, An Open Supply Java-based Framework
The streaming firm works intently with the Spring Boot staff, an open-source Java-based framework. Bakker defined that Netflix makes use of customary Spring Boot plus a curated set of modules that hook into Netflix’s infrastructure. There’s safety integration, dynamic configuration, observability, and extra.
He additional defined that the migration from an older in-house framework to Spring Boot was an extended, deliberate effort. 1000’s of providers have been ported, helped alongside by automated tooling and migration guides. Now, most providers use Spring Boot 3, and upgrades occur inside days of every new launch.
They’ve labored intently with the Spring staff to form the GraphQL assist, and are cautious in adopting Venture Leyden to enhance the backend’s efficiency within the close to future.
Bakker highlights that Spring Boot might not be flashy, nevertheless it’s acquainted, battle-tested, and extensible—precisely what Netflix wants.
Goodbye to Reactive Programming
It might shock some that Netflix, as soon as the birthplace of RxJava, is now transferring away from reactive programming. “Actually each API was once Rx,” Bakker recalled. However over time, the complexity of debugging and sustaining reactive techniques started to outweigh their benefits.
“My sizzling take is that digital threads mixed with structured concurrency goes to fully change reactive programming,” Bakker mentioned.
As we speak, Netflix solely makes use of reactive libraries the place completely vital. For all the things else, the stack has been standardised round synchronous code powered by digital threads.
As structured concurrency solidifies in upcoming Java variations, Netflix sees even much less motive to maintain reactive code round. “We will lastly stay our lives fortunately,” Bakker joked.
The publish The Java Secret of Netflix appeared first on Analytics India Journal.
