Apache Kafka is a widely used distributed streaming platform that thousands of companies rely on to build scalable and high-throughput real-time streaming systems. To operate these systems effectively, it's essential to understand the architecture, key terms, and best practices for working with Kafka. The platform provides scalability, low latency, high throughput, fault-tolerance, flexibility, durability, and real-time data processing capabilities. However, as the system grows in scale, complexity can arise, making it challenging to manage data streams and messages. To address this, New Relic has compiled 20 best practices for operating scalable Kafka clusters, which are categorized into four main areas: working with topics, consumers, producers, and brokers. These best practices cover essential aspects such as partitioning, consumer lag, producer buffer sizes, broker memory, and monitoring, to ensure optimal performance and reliability in Kafka systems.