EventBridge VS Kinesis(Event Stream)

Check out What are event-driven architectures

The key differences between AWS Event Router (EventBridge) and Event Streams (Kinesis) are:

Here's a detailed comparison between Amazon EventBridge and Amazon Kinesis:

Feature	Amazon EventBridge	Amazon Kinesis
Primary Purpose	Event routing and serverless event bus	Real-time data streaming and processing
Data Retention	No built-in retention	24 hours to 365 days
Maximum Record Size	256KB	1MB
Ordering	No guaranteed ordering	Ordered within same partition key
Throughput	Soft limit of 10,000 events/sec per region	Up to 1MB/sec per shard or unlimited with on-demand
Scaling	Automatic	Manual (adding/removing shards) or on-demand
Cost Model	Pay per event	Pay per shard-hour or on-demand capacity
Processing Model	Push-based	Pull-based
Replay Capability	Yes, through archiving	Yes, using sequence numbers
Fan-out Support	Yes, multiple targets per rule	Yes, multiple consumers per stream
Real-time Analytics	No built-in analytics	Yes, with Kinesis Data Analytics
Data Transformation	Yes, via input transformer	Only with additional processing
Cross-Region	Yes	Yes, with replication
Cross-Account	Yes	Yes
Use Cases	• Serverless event routing • Application integration • SaaS integration	• Log aggregation • Real-time metrics • IoT data streaming • Click stream analysis
Latency	Milliseconds	Sub-second
Consumption Pattern	Push to targets	Pull via GetRecords or Enhanced Fan-Out

Event Router (EventBridge)

• Acts like a serverless event bus for routing events between AWS services, SaaS apps, and custom applications

• Best for event-driven architectures where you need to connect multiple services based on rules/patterns

• Supports complex routing rules and filtering

• Pay per event, no capacity planning needed

• Lower throughput (max 10,000 events/sec per account)

• Maximum event size of 256KB

• Near real-time delivery with some latency

Event Streams (Kinesis)

• Designed for real-time streaming data ingestion and processing

• Best for high-throughput streaming use cases like log aggregation, analytics, IoT

• Sequential processing of records in order

• Need to manage capacity through shards

• Much higher throughput (1MB/sec or 1000 records/sec per shard)

• Maximum record size of 1MB

• Sub-second latency

Key considerations for choosing between them:

1. If you need complex routing and integration between services → EventBridge

2. If you need high-throughput sequential processing → Kinesis

3. If cost is primary concern → EventBridge (pay per event)

4. If latency is critical → Kinesis

5. If you need ordering guarantees → Kinesis