AWS Services
Kinesis vs SQS vs EventBridge
Compare Kinesis Data Streams, Amazon SQS, and Amazon EventBridge for streaming logs, durable queues, event routing, replay, ordering, consumers, throughput, and SAA-C03 integration decisions.
After this, you will understand
Streaming questions become less blurry when learners separate a replayable event stream, a durable work queue, and an event router.
Use Kinesis Data Streams for real-time streaming and replay, SQS for worker backlog, and EventBridge for routing events to targets by rules.
Learners use SQS for replayable telemetry streams, use Kinesis for simple job queues, or use EventBridge where consumers need polling and queue-depth scaling.
Classify the event flow: continuous records for many consumers, independent jobs for workers, or routed events from many sources.
Think before readingWhich service fits clickstream data that multiple consumers need to read and replay?
Reading in progress
This page is saved in your local study history so you can continue later.
Study path
Read these in order
Start with the mechanics, then move into the patterns that explain why the system is shaped this way.
Concepts Covered
- Streaming records
- Durable worker queues
- Event routing
- Replay windows
- Shards and partition keys
- Visibility timeout
- Event buses and rules
- Multiple consumers
- Throughput scaling
- SAA-C03 async traps
1. Plain-English Mental Model
Kinesis Data Streams, SQS, and EventBridge all move events or messages, but they represent different communication models.
Kinesis Data Streams = replayable real-time stream
SQS = durable queue for work
EventBridge = event bus and router
Kinesis stores ordered records for a retention window. SQS stores messages until consumers process and delete them. EventBridge matches events with rules and sends them to targets.
The main decision is whether the data is a stream, a job backlog, or a routed event.
2. Why This Service Exists
Some systems produce continuous telemetry: clicks, logs, metrics, device readings, transaction events, or application events. Multiple consumers may need to process the same records and replay from a point in time. Kinesis Data Streams exists for that.
Some systems produce tasks: send email, resize image, process order step, generate report. Workers should process each message and remove it after success. SQS exists for that.
Some systems emit events from AWS services, SaaS providers, or custom apps, and rules decide which targets should react. EventBridge exists for that.
For SAA-C03, look for words such as stream, replay, shard, queue, visibility timeout, event pattern, AWS service event, or multiple independent consumers.
3. The Naive Approach And Where It Breaks
The naive approach is to use SQS for every asynchronous workload.
SQS is excellent for workers, but it does not behave like a replayable ordered stream for multiple independent consumers. If two consumers read from the same queue, they compete for messages.
Another naive approach is to choose Kinesis whenever data is high volume. That can be wrong if the work is simply independent jobs that need backlog and DLQs.
Another mistake is using EventBridge as if it were a worker queue. EventBridge can route events to SQS, but SQS is the component that stores backlog for polling workers.
4. Core Primitives
Kinesis Data Streams primitives include streams, records, shards, partition keys, sequence numbers, producers, consumers, retention, enhanced fan-out, and capacity modes.
SQS primitives include queues, messages, producers, consumers, visibility timeout, dead-letter queues, standard queues, FIFO queues, receive counts, and long polling.
EventBridge primitives include event buses, events, event patterns, rules, targets, Scheduler, Pipes, archives, and replay.
The primitives reveal the purpose. Shards imply streaming scale and order. Visibility timeout implies worker processing. Rules imply event routing.
5. Architecture Use Cases
Use Kinesis Data Streams for clickstream analytics:
web clients -> Kinesis Data Streams -> real-time consumers and storage pipeline
Use SQS for background workers:
API -> SQS -> Lambda or ECS workers
Use EventBridge for event routing:
AWS events + app events + SaaS events -> event bus -> rules -> targets
Use EventBridge to route selected events to Kinesis or SQS when needed. These services are often combined rather than mutually exclusive.
Use Kinesis Data Firehose when the goal is delivery to destinations such as S3, Redshift, or OpenSearch with less consumer code.
7. Security Model
All three services use IAM for API access.
Kinesis producers and consumers need permissions to write and read streams. Enhanced fan-out consumers need registration and read permissions.
SQS producers and consumers need send, receive, delete, and visibility permissions. Queue policies support cross-account and SNS delivery.
EventBridge producers need permission to put events. Event bus resource policies support cross-account patterns. Rules need permission to invoke targets.
KMS encryption can add key-policy requirements. Event and message payloads should not contain sensitive data unless every consumer or target is authorized.
8. Reliability And Resilience
Kinesis stores records for a retention period, so consumers can recover by replaying while records remain available.
SQS preserves messages through worker failure until they are deleted or expire. DLQs isolate poison messages.
EventBridge retries target delivery and can use DLQs for supported targets. Archives and replay can help recover event-routing flows.
Consumers should be idempotent. Duplicate processing can happen through retries, producer retries, or failure recovery.
Retention windows matter. Kinesis records expire after retention. SQS messages expire after retention. EventBridge archives must be configured if replay is required.
9. Performance And Scaling
Kinesis performance is shaped by shards, partition keys, producer batching, consumer throughput, enhanced fan-out, and capacity mode.
SQS standard queues scale for high-throughput work queues. FIFO queues preserve ordering within message groups with different scaling characteristics.
EventBridge scales event routing, but target capacity can still throttle. Precise event patterns reduce unnecessary invocations.
If many consumers need the same stream with independent high throughput, Kinesis enhanced fan-out is a strong signal.
If workers scale based on queue depth and message age, SQS is a strong signal.
10. Cost Model
Kinesis cost depends on shards or on-demand throughput, PUT payload units, consumers, enhanced fan-out, and retention.
SQS cost is request-based and can be optimized with long polling and batching.
EventBridge cost depends on events, matched rules, API destinations, archives, replay, Scheduler, and Pipes.
The wrong primitive adds hidden cost: custom replay logic, duplicated Lambda invocations, missed backlog, or unnecessary stream management.
Choose the simplest correct model.
12. SAA-C03 Exam Signals
"Real-time streaming records" points to Kinesis Data Streams.
"Replay records from a retention window" points to Kinesis Data Streams.
"Multiple consumers process the same record stream" points to Kinesis Data Streams.
"Worker queue and visibility timeout" points to SQS.
"Failed work moves to DLQ" often points to SQS.
"Route AWS service events by rules" points to EventBridge.
"SaaS partner event source" points to EventBridge.
13. Common Exam Traps
Do not use SQS when each consumer needs its own copy of a replayable stream.
Do not use Kinesis when a simple durable work queue is enough.
Do not use EventBridge as the backlog mechanism for polling workers.
Do not forget partition key hot spots in Kinesis.
Do not forget visibility timeout in SQS.
Do not forget target permissions in EventBridge.
15. Related Topics
Review Amazon Kinesis Data Streams, Amazon SQS, Amazon EventBridge, SQS vs SNS vs EventBridge, and Event-Driven Order Processing.
Official AWS references:
What to study next
These links keep the session moving: read prerequisites first, then open the systems, concepts, and patterns that deepen this page.
Prerequisites
Read these first if the mechanics feel unfamiliar.
More Links
Additional references connected to this page.