Performance replication shares most of the data between clusters, except for local mounts, tokens, and leases. A performance replication secondary cluster can handle requests and can serve read operations locally, but forwards any operations that modifies data to the primary cluster. You can think of performance replication as a way to horizontally scale your Vault infrastructure.
Disaster recovery replication, as the name implies, is mainly used for disaster recovery scenarios, and replicates all data including token, leases, and local mounts. This is helpful in catastrophic failure scenarios where a primary cluster becomes unavailable. A DR secondary cluster can be promoted as the new primary and continue serving requests with little downtime. However, unlike performance replication secondary clusters, DR secondary clusters are unable to serve requests until they get promoted as the primary.
This page goes into more detail and comparison between the two forms of replication: https://www.vaultproject.io/docs/enterprise/replication/index.html#performance-replication-and-disaster-recovery-dr-replication