WuKongIM Docs

Product Overview

Quickly decide what WuKongIM is, which products it fits, and what your application still owns.

WuKongIM is open-source, self-hosted real-time messaging infrastructure. It lets applications provide chat, notifications, and multi-device message synchronization on their own infrastructure, and is designed for large-scale connection and messaging workloads.

v3 Beta

WuKongIM v3 is still in beta. APIs, configuration, and storage formats may change before the stable release. Before production use, review Upgrade & Migration, Security & Permissions, and Health & Monitoring.

The problem it solves

A real-time messaging product needs more than a way to send data. It must manage many long-lived connections, per-channel message order, persistence, reconnect recovery, and multi-device state. WuKongIM provides these capabilities in one self-hosted cluster so that application teams can focus on their product and business rules.

An application or business service sends messages to a channel. A channel can represent a person, group, or product-specific topic. WuKongIM stores and synchronizes the messages, delivers them to online devices, and lets offline or reconnecting devices catch up later.

Responsibility boundaries

ParticipantPrimary responsibility
WuKongIMClient connections, per-channel message order, persistence and replication, online delivery, offline recovery, and multi-device synchronization
Application backendAccounts, credentials, friend or group relationships, product authorization, content governance, mobile push, and business data
Client SDKConnection lifecycle, local message state, presentation, and user interaction

What it is not

  • It is not a chat application with a user interface and complete product workflows.
  • It does not replace the application backend or decide who may send which messages.
  • It is not a general queue for offline jobs, global transactions, or long-term analytics.

Is it a fit?

WuKongIM is worth evaluating when a product needs many online users or devices, per-channel ordering, reconnect recovery, durable history, or multi-device synchronization. Chat, notifications, and customer service are the most direct fits; live interaction, IoT, and AI messaging workflows can use the same foundation.

If the requirement is only low-frequency background work, offline batch processing, or analytical queries, a task queue, stream processor, or analytical database is usually a better fit.

One deployment model

Every WuKongIM deployment is a cluster. With one node, it is a single-node cluster that follows the same cluster paths as a multi-node deployment; there is no standalone mode that bypasses cluster semantics. Effective capacity depends on hardware, storage, replicas, channel distribution, and workload, so validate it with a representative scenario.

On this page