EventBridge is a fully managed, serverless event data service from Alibaba Cloud that acts as a central hub for data integration and processing in the AI-native era. EventBridge connects application components through events to build loosely coupled and scalable event-driven architectures (EDAs).
This topic describes the core resources, basic concepts, and typical scenarios of EventBridge.
Event-driven architecture
An event-driven architecture is a software design pattern where system components collaborate by publishing and responding to events instead of making direct calls to each other. This pattern provides loose coupling and high scalability. It is ideal for scenarios such as microservice orchestration, real-time data processing, and automated O&M.
For more background information, see:
Core resources
EventBridge is based on the CloudEvents 1.0 standard and connects Alibaba Cloud services, custom applications, and third-party SaaS platforms. Using its built-in filtering, transformation, and routing capabilities, EventBridge delivers events from sources to target systems without requiring integration code.
EventBridge provides three types of core resources for different event processing scenarios:
Dimension | Event bus | Event stream | EventHouse |
Locate | Event routing and distribution | Continuous collection and transmission of massive data | Persistent storage and query analysis of events |
Routing pattern | N:M (many-to-many) | 1:1 (one-to-one) | — |
Core capabilities | Filtering, transformation, and multi-target delivery | High throughput, low latency, and consumption replay | Instant SQL queries and low-cost storage |
Typical scenarios | Microservice decoupling, SaaS integration, and automated O&M | Log collection, IoT data aggregation, and real-time ETL | Event auditing, root cause analysis, and BI reports |
Event bus
An event bus is the core routing resource of EventBridge. It receives events and distributes them to one or more targets.
Based on the CloudEvents 1.0 standard, an event bus accepts events from Alibaba Cloud services, custom applications, or SaaS applications. By configuring event rules, events can be filtered and transformed before they are delivered to target services such as Function Compute, ApsaraMQ for RocketMQ, and DingTalk.
Key capabilities
Content filtering: Filters events based on their content using event rules.
Event transformation: Modifies the event structure during routing to match the data format required by the target system.
Multi-target delivery: Triggers multiple downstream services to process an event from a single publication.
Typical scenarios
Microservice decoupling: An order service publishes an "Order Created" event. The inventory, logistics, and rewards services subscribe to the event independently, without being aware of each other.
SaaS integration: When customer information is updated in Salesforce, it is automatically synchronized to an internal CRM system.
Automated O&M: When Cloud Monitor detects an instance failure event, it automatically triggers an O&M script or sends an alert notification.
Event stream
An event stream is a high-throughput data transmission resource in EventBridge. It is responsible for the continuous collection and real-time processing of massive amounts of data.
Unlike the many-to-many routing of an event bus, an event stream provides a one-to-one (1:1) data transmission service designed for high-throughput and ordered data scenarios. It is suitable for processing continuous data streams such as logs, monitoring metrics, and user behavior data.
Key capabilities
High throughput and low latency: Supports data writes and reads at millions of transactions per second (TPS).
Consumption replay: Supports pulling data from a specific point in the past for reprocessing.
Typical scenarios
Real-time data warehouse: Collect binary logs (binlogs) from business databases or application logs in real time, process them with Flink, and then write them to a data warehouse.
IoT data collection: Aggregate status data reported by a massive number of devices through an event stream and then distribute it to downstream analytics systems.
Clickstream analysis: Capture user behavior data on websites or apps in real time for use by recommendation systems.
Event house
An event house is a structured event storage and analysis resource in EventBridge. When an event house is enabled, all event data that passes through EventBridge is stored persistently and can be queried and analyzed at any time.
An event house uses a columnar and tiered storage architecture to reduce storage costs while ensuring query performance. In addition to retaining event history for traceability and disaster recovery, it supports instant SQL queries to transform event data into business insights.
Key capabilities
Instant SQL queries: Allows you to run standard SQL queries directly on historical events without moving the data to a data warehouse. This capability supports multi-dimensional aggregation, filtering, and analysis.
Low-cost lakehouse storage: Uses a tiered storage architecture suitable for long-term retention of massive amounts of event data.
Typical scenarios
Intelligent O&M and root cause analysis: When a system fails, you can use SQL to quickly retrieve the chain of anomalous events within the failure time window to locate the root cause.
Business intelligence reports: You can connect tools such as Quick BI to an event house to generate real-time business reports based on event streams.
Event audit and compliance: You can retain business event records for the long term to meet audit and compliance requirements.
How to choose
Choose the appropriate resource based on your scenario:
To route events from multiple sources to multiple targets, use an event bus. For example, an order event needs to simultaneously notify the inventory, logistics, and rewards systems.
To build a high-throughput data pipeline between two systems, use an event stream. For example, continuously synchronize database change logs to a downstream analytics system.
To store and query historical event data, use an event house. For example, analyze the distribution of all failed payment events over the past month.
Note: These three resources can be used together. For example, an event bus can receive and route events, an event stream can provide a high-throughput channel, and an event house can store historical data for query and analysis.
AI scenarios
The built-in event filtering and transformation engine in EventBridge performs real-time traffic scrubbing and feature enhancement before the data enters an AI model. It also supports routing the processed events to a machine learning platform or a vector database.
Scenario | Resource used | Description |
AI real-time inference trigger | Event bus | When OSS receives an image upload event, it automatically triggers Function Compute to call an AI model for recognition. |
AI training data pipeline | Event stream | Continuously collect user behavior data to serve as a real-time input source for model fine-tuning. |
AI agent context | Event Store | Serves as a knowledge source for retrieval-augmented generation (RAG), allowing an AI agent to query historical business events. |
Terms
Concepts | Description |
Event | A data record of a system state change. It is the basic unit that EventBridge processes. |
Event source | The source of an event. It is responsible for producing events. An event source can be an Alibaba Cloud service, a custom application, or a SaaS platform. |
Event target | The processing endpoint for an event. It is responsible for consuming events. An event target can be Function Compute, ApsaraMQ for RocketMQ, an HTTP endpoint, or another service. |
Event bus | The routing hub for events. It is responsible for receiving, filtering, transforming, and distributing events. |
Event rule | Defines the conditions for matching events and the delivery targets. When a matching event occurs, the event is routed to the associated event targets. |
For more information, see Terms.
Benefits
Billing
For more information, see Billing overview.