No tutorials. Hard-won operational understanding.
Distributed systems, Kafka, Kubernetes, JVM performance and the parts that only become obvious under production pressure.
Engineering note
Why Your Service Mesh Is Slower Than You Think
The actual latency tax inside sidecar proxies: connection churn, TLS handshakes and filter pipelines.
Engineering note
The Backpressure Problem No One Talks About
Why Kafka consumer lag is a symptom, not a mechanism, and how async boundaries silently break downstream pipelines.
Engineering note
Kubernetes StatefulSet Gotchas in Production
Ordered rollouts, PVC lifecycle, local PV races and node drains — the StatefulSet behavior that matters under real operations.
Engineering note
Raft Is Not Magic: What the Paper Does Not Explain
The operational gaps around pre-vote, leader leases, snapshots, partitions and log divergence in production Raft.
Engineering note
GC Tuning Is Not About the GC
Why profiling allocation rates and object lifetimes beats blindly tweaking JVM collector flags.
Engineering note
The Real Cost of Exactly-Once in Kafka
Transactions, producer fencing, read_committed and the latency tax behind stronger guarantees.