Abhishek Singh

In Kafka, we have topics. Producers send data to a topic, and consumers pull data from it. Each topic has multiple partitions. You might have heard your team lead say that we can simply increase the partition count to make our cluster more scalable. There is no ordering guarantee across partitions for a topic, but messages within a partition are ordered. This can be confusing initially, but think of consumer groups as subscribers: each group has multiple consumers pulling data from a topic, and the group maintains its own offset for that topic. Check out the pinned post on my profile to understand this in detail!