Zookeeper

Distributed coordination service for distributed applications that exposes a small set of primitives for synchronization, configuration maintenance, naming, and group membership. It relieves distributed systems from implementing coordination logic from scratch, which is notoriously error-prone due to race conditions and deadlocks.

Apache ZooKeeper has been used in production at Yahoo, Netflix, and many enterprises since 2008. It underpins Apache Kafka, Hadoop, and other systems that rely on distributed coordination. Its hierarchical namespace and simple API keep complexity low, while its replicated design and in-memory data store deliver high throughput and low latency for read-heavy workloads.

Core capabilities:

• Hierarchical data model, similar to a file system, with znodes for configuration, status, and coordination data
• Distributed synchronization primitives: leader election, distributed locks, barriers
• Watches for change notifications when znodes are updated, so clients can react without polling
• Replicated ensemble that stays available as long as a majority of servers are up
• Sequential consistency, atomicity, and single-system view across all clients

Use cases include service discovery and registration, distributed configuration management, leader election for distributed systems, and coordination for message brokers and data pipelines. Developers integrate it via Java or C bindings, often as the backbone for systems like Kafka that need reliable cluster metadata and failure recovery.