May 4 · 10 min read · Source: Reasons Java services get slower after a few hours: How to find thread pool saturation? You boot a service at 03:00 and it behaves beautifully: sub-second responses, healthy throughput. By noon, requests start lagging. After an outage ticke...
Join discussion
May 3 · 21 min read · TLDR: Java 21 LTS makes virtual threads a production-ready replacement for bounded thread pools — your newFixedThreadPool(200) can become newVirtualThreadPerTaskExecutor() and handle 10× the concurrency with no architectural changes. Pattern switch w...
Join discussionMay 3 · 24 min read · TLDR: Java 14–17 ran a deliberate four-release preview-to-stable conveyor belt. Records replaced 50-line POJOs with one line. Text blocks ended escape-sequence chaos in multi-line strings. Sealed classes turned "please only subclass these types" comm...
Join discussionMay 3 · 15 min read · TLDR: Java 8 introduced the most impactful set of language features in Java's history — lambdas eliminated anonymous inner classes, streams replaced imperative loops, and Optional made null handling explicit. Java 9's module system drew a boundary ar...
Join discussionApr 25 · 12 min read · Source: Methods to Tune JVM Container Awareness for Real Kubernetes Resource Constraints You shipped the service, the green pipeline smiled, and then production nodes started killing pods. The symptom is deceptively simple — an OOMKilled pod — but ...
Join discussion
Apr 24 · 13 min read · Source: Methods to Handle Time Drift Issues in Distributed Systems Many production incidents start with a moment of cognitive dissonance: logs from service A show an event happening "after" an event in service B, but tracing reveals the opposite. D...
Join discussion
Apr 19 · 33 min read · TLDR: Spark executor OOMs are almost never caused by insufficient total cluster RAM — they are caused by misallocating memory across five distinct JVM regions while ignoring GC behavior and memoryOverhead. Master the UnifiedMemoryManager model, apply...
Join discussionApr 16 · 10 min read · Source: Reasons Thread Pool Misconfiguration Creates Fake Throughput in Java Backends 1. A quiet alarm: why you shouldn't trust requests/sec alone In a production incident I investigated, dashboards showed impressive requests/sec during a sudd...
Join discussion
Apr 10 · 24 min read · TLDR: JVM garbage collection automatically reclaims unused heap memory, but every algorithm makes a different trade-off between throughput, latency, and memory footprint. The default G1GC targets 200ms pause goals and works well for most services. Fo...
Join discussion
