I’m curious what you actually run in production—vote in the poll and tell us why in the comments (JDK version, heap size, latency/throughput goals, containerized or not).
Quick refresher:
- G1 → solid default since JDK 9/11; balanced pauses, easy ops.
- ZGC → sub-10ms pauses at multi-GB/-TB heaps; great for low-latency services.
- Shenandoah → concurrent compaction, predictable pauses; strong on large, active heaps.
- Parallel → max throughput when pauses are acceptable; batch jobs & ETL friendly.
If you’re testing a switch: run A/B canaries, compare throughput, tail latency, CPU%, GC cycles, and allocation rate under realistic load.
Cast your vote: G1 / ZGC / Shenandoah / Parallel — and share your context to help others choose wisely. 🙌
#Java #GarbageCollector #JVM #Performance #Latency #Throughput #G1GC #ZGC #Shenandoah #ParallelGC #ProdReady #SRE #DevOps #Poll
#garbagecollector #java #poll
Giorgi said:
most of engineers don't even think whether it's default or whatever, we/they just know there's GC happening, and that's it.