What Is ZooKeeper? Core Concepts, Node Types, and Quick Start Guide

Introduction to ZooKeeper

ZooKeeper is a high-performance coordination service designed for distributed applications. It encapsulates common services such as naming, configuration management, synchronization, and group management into simple interfaces, allowing developers to implement consensus, group management, leader election, and presence protocols without building them from scratch.

Data Model and Znode Types

ZooKeeper maintains a hierarchical data model similar to a file system, but stores data in memory for faster access. Each node in this hierarchy is called a znode and can hold data (typically small pieces of state, configuration, or location information). Znodes come in four types:

PERSISTENT : The most common type. The node persists after the client disconnects from ZooKeeper.

PERSISTENT_SEQUENTIAL : A persistent node to which ZooKeeper appends a monotonically increasing sequence number to the node name.

EPHEMERAL : The node is deleted when the client session ends. Ephemeral nodes cannot have children.

EPHEMERAL_SEQUENTIAL : An ephemeral node that also receives a sequential number suffix.

The article includes a diagram illustrating the znode tree structure.

Watch Mechanism

ZooKeeper provides a watch (listener) mechanism. Clients can register watches on znodes; when a znode's data changes, ZooKeeper notifies the watching clients. This pattern resembles the Observer design pattern and is fundamental for implementing configuration management, state synchronization, and other distributed primitives.

Installation and Basic Operations (Version 3.8.3)

The article walks through a minimal installation on a Linux machine with Java pre-installed:

tar -zxvf apache-zookeeper-3.8.3-bin.tar.gz
cd apache-zookeeper-3.8.3-bin
cp conf/zoo_sample.cfg conf/zoo.cfg
bin/zkServer.sh start

After starting the server, the CLI client ( bin/zkCli.sh) is used to verify the service: ls / lists the root znode (initially only zookeeper). create /fullstack followit creates a new persistent znode with data "followit". ls / again shows the newly created fullstack node. get -s /fullstack retrieves the node's data and metadata (including version numbers and timestamps).

Screenshots of the CLI output are included to demonstrate each step.

The article concludes by noting that this is an introductory overview and promises deeper articles in the future.