Raft is a consensus algorithm designed to keep distributed systems running smoothly by ensuring that multiple servers remain in sync. Unlike traditional systems that rely on a single machine, Raft addresses the issue of server failures and data loss. It does this by organizing servers into replicated state machines, which consist of a state machine, a log, and a consensus module. The state machine produces consistent outputs when given the same commands in the same order. The consensus module ensures that all servers have the same log entries, allowing them to maintain coherence even if some servers fail. In Raft, one server is elected as the leader, responsible for managing log replication to follower servers. This leader handles requests, stores entries in its log, and sends these entries to followers using AppendEntries messages. The algorithm checks for various scenarios, such as missing or inconsistent log entries on followers, and adjusts by sending the necessary data to achieve consistency. A key aspect of Raft is its commitment process: the leader requires that a majority of servers replicate a log entry before it is considered committed. Once committed, an entry cannot be overwritten by future leaders. Raft also prevents multiple leaders from arising in the system. If a leader fails to send heartbeat messages, followers can timeout and initiate an election to choose a new leader. Candidates for leadership must receive a majority of votes in order to take over. This structured approach ensures that there is always one leader at a time, maintaining stability in the distributed environment. The article breaks down these concepts in a straightforward manner, making them accessible to readers without a deep technical background.