How do you redistribute data across a cluster when changing the number of partitions?

When the number of partitions changes, the cluster must rebalance data to maintain even distribution and performance. The process typically involves: 1) Determining the new partition layout and mapping of data ranges to nodes. 2) Initiating a rebalancing operation (e.g., Kafka's partition reassignment tool or Cassandra's nodetool move). 3) Streaming data from old partitions to new ones, ensuring consistency by using write-ahead logs or replication. 4) Updating metadata so that clients route to the correct nodes. 5) Monitoring progress and handling failures by retrying or rolling back. The goal is to minimize downtime, avoid data loss, and keep the cluster healthy during the transition.