In Part 2, we implemented leader election and got our cluster to choose a leader and maintain leadership through heartbeats. But a leader that can’t do anything useful isn’t much of a leader! Now it’s time to implement the heart... Read more about Implementing Raft: Part 3 - Log Replication
In the previous post, we explored the theory behind Raft’s leader election and log replication processes. We learned how nodes transition between follower, candidate, and leader states, and how they use terms and voting to reach consensus on who should... Read more about Implementing Raft: Part 2 - Leader Election
There is a good chance that you use systems using the Raft algorithm, or a modified version of it, every day. It’s the engine for consensus in systems like etcd (the backbone of Kubernetes), Kafka (with KRaft), Redis (with RedisRaft),... Read more about Implementing Raft: Part 1 - The Why and How of Consensus in Distributed Systems
I was recently reading about how Cassandra uses a Bloom filter to check if any of its SSTables are likely to contain a given partition key or not without having to read the entire SSTable. This got me thinking about... Read more about Let's Build a Bloom Filter in Ruby
Transport Layer Security (TLS) secures most of today’s internet traffic by encrypting data in transit to create secure communication channels. While its encryption is often celebrated for privacy, one of its most critical roles is preventing unintended interference from intermediaries,... Read more about How TLS Neutralizes Intermediaries on the Web
Speed is a feature, and in fact, for some applications it is the feature. — Ilya Grigorik - High Performance Browser Networking Like everything in software engineering, every decision involves trade-offs. TCP’s Nagle’s algorithm is a perfect example of balancing... Read more about Batch or Burst? Disabling Nagle’s Algorithm with TCP_NODELAY for Better Latency
This article will explore how to build a Change Data Capture (CDC) pipeline to capture data changes in a PostgreSQL database and propagate those changes in real-time using Debezium, AWS Kinesis, and EventBridge. We will cover the fundamental concepts of... Read more about Turn Data Changes into Events with Debezium and EventBridge
This article explains how to build a Kubernetes admission controller webhook that automatically injects sidecar containers into pods based on specific annotations or labels. Understanding the sidecar pattern A sidecar container is a container that runs alongside the primary application... Read more about Extending Kubernetes: How to Automate Sidecar Injection with Admission Controller Webhooks
Node.js is traditionally known as a single-threaded, asynchronous, event-driven JavaScript runtime. It’s designed for building scalable network applications using non-blocking I/O and an event-driven asynchronous paradigm rather than relying on multithreading. Before diving into the details, let’s differentiate between the... Read more about Node.js Multithreading!