ZooKeeper is a widely used distributed coordination and synchronization service for distributed applications. It relies on the built-in atomic broadcast protocol Zab to provide high availability and guarantees such as totally ordered updates for the overlying applications. ZooKeeper’s flexibility, however, is limited with regard to the underlying Zab protocol.In this work, we propose to switch ZooKeeper’s underlying consensus protocol at runtime while preserving all guarantees for the overlying applications. We present an alternative protocol Instant Commit, which is write delay optimized. We implemented the adaptation logic and Instant Commit in ZooKeeper and show its potential with a workload-aware adaptation trigger.