This article provides a standardized JVM tuning methodology with three production-ready parameter templates for low (2C4G), mid (8C16G), and high (16C32G+) spec servers, covering G1 and ZGC configurations, common pitfalls, and verification steps.
The article explains why a large product cache moved from an in‑heap Caffeine cache to an off‑heap OHC solution, detailing serialization trade‑offs, per‑dimension cache configuration, singleflight loading, and the performance gains observed during high‑traffic promotions.
The article examines why Java 26 has sparked intense discussion—not for flashy syntax but for concrete upgrades like HTTP/3, performance tweaks, and GC improvements—guiding developers on when and how to adopt the new version responsibly.
This article details a full-chain optimization of a Flink CDC real-time data warehouse pipeline syncing 20 MySQL instances to StarRocks via Kafka, resolving GC crashes, checkpoint failures, and 30-minute latency through resource scaling, parallelism redesign, memory tuning, checkpoint configuration, and startup strategy changes, achieving 5-second end-to-end latency and 100% checkpoint success.
Xiaohongshu’s middleware team migrated thousands of Java services from JDK 8 to RedJDK 11/17, achieving over 10% performance gains, 50% GC pause reduction, and eliminating OOM crashes through systematic JDK upgrades, GC tuning, native‑memory improvements, and standardized deployment pipelines.
By analyzing heap dumps with MAT and scrutinizing GC logs, the Vivo Internet Server Team identified thread‑local FutureAdapter and Jackson BufferRecycler objects and tuned promotion thresholds, cutting the Java game service’s Full GC occurrences from roughly 120 per day to about 30 and dramatically shortening pause times.
The article details how a Go service’s default GOGC setting caused overly frequent garbage‑collection pauses that spiked request timeouts, and how adjusting GOGC dynamically with debug.SetGCPercent and setting memory limits reduced GC CPU usage, extended pause intervals, and eliminated timeout spikes.
This article explains how Kafka achieves million‑message‑per‑second throughput by leveraging zero‑copy I/O, an append‑only log, batch processing, compression, consumer pull optimization, unflushed memory buffers, and JVM garbage‑collection tuning, detailing each mechanism and its impact on performance.
Didi upgraded its self‑developed Elasticsearch from JDK 11/G1 to JDK 17, adopting ZGC for latency‑critical clusters and tuned G1 for throughput, which eliminated long GC pauses, reduced query latency by up to 96%, cut CPU usage, and dramatically improved stability across multiple production clusters.
This article provides a comprehensive guide on optimizing Spark and Flink workloads, covering parallelism settings, garbage‑collection tuning, out‑of‑memory mitigation, and full production‑grade configuration examples for both frameworks.
This article details how JD.com tackled rapid growth of its flash‑sale product pool by analyzing JVM GC bottlenecks, redesigning the system architecture, introducing dual‑cache bucket updates, local LRU caching, Bloom filters, and service separation, ultimately achieving significant performance and stability improvements for large‑scale promotions.
Long Young GC pauses in gRPC/Netty services were traced to Netty’s default thread‑local cache creating many MpscArrayQueue objects, and disabling the cache with the JVM options ‑Dio.netty.allocator.useCacheForAllThreads=false and ‑Dio.grpc.netty.shaded.io.netty.allocator.useCacheForAllThreads=false reduced GC time from up to 900 ms to about 100 ms, stabilizing the service.
This article presents a performance comparison of Dubbo 2, Dubbo 3 interface‑level, and Dubbo 3 application‑level address discovery models under a million‑instance scenario, showing significant reductions in memory usage and GC frequency for the newer models.
This article explains how Kafka’s architecture, use of OS page cache, Sendfile optimization, and a custom memory buffer pool work together to minimize JVM garbage collection overhead and deliver the massive throughput required by big‑data messaging workloads.
This article explains how Kafka’s design—its broker architecture, use of sequential disk I/O, PageCache, Sendfile, and a custom memory buffer pool—optimizes JVM garbage collection and achieves massive throughput in big‑data messaging scenarios.
This article details a series of practical enhancements applied to an HBase 1.2.4‑based cluster—including layered BucketCache, data pre‑heating, GC‑friendly object pooling, and a multiget concurrency model—that together raise throughput several‑fold and consistently keep P99 latency below 50 ms in YCSB benchmarks.
Long garbage collection pauses can degrade Java application performance and SLA, so this guide outlines seven practical steps—including reducing object allocation rates, increasing young generation size, selecting appropriate GC algorithms, avoiding swap usage, tuning GC threads, managing I/O load, and limiting System.gc() calls—to minimize pause times.
This article outlines JD's attempts and explorations in customizing the JDK for HDFS, describing the background, limitations of Oracle JDK 1.8, the adoption of OpenJDK 11 with G1GC, a series of JVM and GC optimizations, performance results, and future development plans.
This article explains how Alibaba engineers reduced young‑generation garbage‑collection pauses in large‑scale HBase deployments from over a hundred milliseconds to just a few milliseconds by redesigning memory management with CCSMap, BucketCache, and the tenant‑aware AliGC algorithm.
The article presents Pitcher, a Go‑implemented reverse‑proxy middleware, detailing its one‑core‑multiple‑routes architecture, callback‑based module model, hot‑configuration reload, graceful restart strategies, GC optimizations, service splitting, and various transport choices such as TCP/protobuf, UDP, and Unix domain sockets to achieve high‑performance, scalable traffic entry for Qunar.
