PinnedInLevel Up CodingbyRadhakishan Surwase·Oct 28, 2024Mastering Real-Time Data Pipelines in Golang with Channels and Worker PoolsProcessing a high volume of messages in real-time is a common challenge for systems interacting with event-driven queues, like Kafka or…
InLevel Up CodingbyRadhakishan Surwase·Oct 680,000 Devices. 13 Rules. Zero Data Lost.What 80,000 GPS devices and 691 million pings a day teach you about building systems that never drop data.A response icon1A response icon1
Radhakishan Surwase·Sep 7Zero Data Loss in Kafka-Backed SEDA: A Sequencing Rule, Not a SettingReplication factors and acks=all are necessary. They are not sufficient — the guarantee is only as strong as the order in which a stage…
InLevel Up CodingbyRadhakishan Surwase·Aug 27Fan-Out, Not Fork-Lift: Engineering Geospatial Intelligence into a Go GPS Pipeline with…Every GPS platform eventually hits the same wall. The time-series store you built the system on — reliable, well-indexed, battle-tested for…
InLevel Up CodingbyRadhakishan Surwase·Aug 24Two Customers, One Item, Same Second: Solving Inventory Oversell Under ConcurrencyProblem StatementA response icon1A response icon1
InLevel Up CodingbyRadhakishan Surwase·Aug 18Kafka Isn’t a Message Broker. It’s Five Different Architectures Wearing One Name.Ask most engineers what Kafka does, and you’ll get some version of “it’s a message queue, but faster and it keeps the messages around…A response icon1A response icon1
InLevel Up CodingbyRadhakishan Surwase·Aug 17A Driver’s Phone Loses Signal for 40 Minutes — Then Reconnects With 480 Queued GPS PingsDo the math on the reconnect, not just the outage. A driver enters a dead zone — common on rural highways, common enough that it’s a daily…
InLevel Up CodingbyRadhakishan Surwase·Aug 13Distributed Locks in Go: Why Fencing Tokens MatterImagine a ticketing system with one concert ticket left.Alice and Bob click Buy at exactly the same time. Two application servers handle…
InLevel Up CodingbyRadhakishan Surwase·Aug 13The Architecture of Failure: Building Fault-Tolerant Go Microservices That Refuse to DieGo powers Docker, Kubernetes, and some of the highest-throughput API gateways in production today. But at cloud scale, fault tolerance was…
InLevel Up CodingbyRadhakishan Surwase·Aug 12Mastering Exactly-Once Semantics in Go Microservices: The Myth, the Math, and the ArchitectureNetworks drop packets, processes crash mid-write, and true exactly-once delivery is mathematically impossible. This piece breaks down how…