API Latency Issue Explained in 30 seconds

[00:00:00]Let's talk about a real-life back-end engineering problem. Why is my API fast in testing but slow in production? This is one of the most misleading back-end issues. In local testing or staging environments, APIs look very fast, but in production, the same API becomes slow. The reason is network latency and distributed dependencies. In production, a single API call might involve authentication service call, database query, cache lookup, third-party API call, logging service, and much more.

[00:00:34]Now, each network hop adds latency. Even if each service is fast individually, the combined latency becomes significant. In one real case, an API took 50 ms locally, but over 1.2 seconds in production. The fix was optimization of service calls. So, in our case, we reduced the unnecessary synchronous calls [music] and introduced batching.

[00:00:57]We also moved non-critical operations like logging and notifications to asynchronous processing using Kafka. So, what's the lesson? Production systems are slow not because of one service, but because of too many network hops. If you want to know in detail about the basics of Kafka for asynchronous communication, refer to this 3-hour crash course video, and make sure you subscribe to the channel.