§ Journal
Systems, explained until they can be tested.
Long-form articles for engineers. Each one carries working calculators and simulators so assumptions can be changed rather than taken on trust. RSS.
Scaling WebSockets and SSE to Hundreds of Pods: Where Redis Pub/Sub Breaks and How to Fix It
Why realtime fan-out over Redis Pub/Sub degrades as a WebSocket/SSE gateway grows to hundreds of pods: O(pods) delivery, cluster broadcast, slow-subscriber disconnects, reconnect storms, and the fixes: sharded Pub/Sub, sequence numbers with replay, room affinity, backpressure.
From MySQL to Elasticsearch: Building a Reliable CDC Pipeline That Scales to 1,000 RPS
A practical design for streaming MySQL changes into Elasticsearch with Debezium and Kafka: ordering, idempotency, failure recovery, and an honest derivation of what 1,000 RPS means for CDC event rates.
Designing a Real-Time IoT Data Platform: MQTT, Kafka, ksqlDB and MongoDB
How device telemetry travels from MQTT through Kafka and ksqlDB into MongoDB: QoS semantics, partitioning, event-time windows, time-series storage, and the failure modes in between.
Inside an IP Camera: Understanding RTSP, Codecs, FPS, I-Frames and Video Streaming
How an IP camera turns photons into an RTSP stream, what H.264/H.265, GOP and bitrate settings actually change, and how those choices propagate to VMS capacity, latency and ANPR or object-detection accuracy.
Kafka as a Distributed Log: Partitions, Replication, Consumer Groups and Replay
A systems-level guide to Kafka topics, partition ordering, producer acknowledgements, replication, consumer offsets, retention and replay, with interactive partition and consumer-group playgrounds.
RabbitMQ Beyond the Queue: Exchanges, Priorities, Persistence and Delayed Work
A practical guide to RabbitMQ routing with direct, topic and fanout exchanges, priority delivery, durable queues, message TTL and delayed retries with dead-letter exchanges.
Redis Beyond the Cache: Persistence, Pub/Sub and Streams, and What Each Actually Guarantees
What Redis really promises once it holds data you cannot lose: RDB and AOF durability, async replication and failover loss, Pub/Sub at-most-once, lists as queues, and Streams consumer groups compared with Kafka.