25 editable reference diagrams
Streaming Architecture Diagrams
Streaming architectures process records while producers and consumers operate independently. These diagrams make brokers, partitions, stateful processing, delivery failures and downstream sinks visible in one design conversation.
Define the ordering scope and acceptable processing delay before selecting a topology. Repeated delivery, consumer lag and replay are operational realities. Annotate event keys, state checkpoints and side-effect idempotency rather than inferring end-to-end guarantees from the broker alone.
Choose a diagram for your design question
Open a reference to inspect its component flow and trade-offs, then adapt it in the browser-local editor. Manual editing does not require login; AI generation and account storage do.

Apache Kafka architecture
Ordering is scoped to a partition.
6 components · 5 connections
Explore apache kafka architecture →
Kafka Event-Driven Microservices Architecture
Asynchronous consumers
6 components · 6 connections
Explore kafka event-driven microservices architecture →
Kafka producer consumer architecture
Commit offsets after safe side effects.
6 components · 5 connections
Explore kafka producer consumer architecture →
Kafka Connect architecture
Connector offsets need durable management.
7 components · 5 connections
Explore kafka connect architecture →
Kafka Streams architecture
Understand state recovery and repartition cost.
6 components · 5 connections
Explore kafka streams architecture →
Kafka Schema Registry architecture
Test compatibility before publishing schema changes.
8 components · 6 connections
Explore kafka schema registry architecture →
Kafka multi cluster replication architecture
Define offset translation before failover.
6 components · 5 connections
Explore kafka multi cluster replication architecture →
Kafka dead letter queue architecture
Replay must not repeat unsafe effects.
7 components · 7 connections
Explore kafka dead letter queue architecture →
Apache Flink architecture
Checkpoint consistency includes source and sink behavior.
7 components · 5 connections
Explore apache flink architecture →
Spark Structured Streaming architecture
Validate sink behavior under replay.
5 components · 4 connections
Explore spark structured streaming architecture →
Apache Pulsar architecture
Separate broker serving from durable storage.
6 components · 5 connections
Explore apache pulsar architecture →
RabbitMQ architecture
Choose acknowledgement and routing policies deliberately.
7 components · 6 connections
Explore rabbitmq architecture →
NATS messaging architecture
Distinguish transient messaging from durable streams.
6 components · 5 connections
Explore nats messaging architecture →
MQTT messaging architecture
Choose session and delivery settings per device.
5 components · 4 connections
Explore mqtt messaging architecture →
Redis Streams architecture
Inspect and recover abandoned pending work.
6 components · 5 connections
Explore redis streams architecture →
Event streaming platform architecture
Treat event schemas as shared product contracts.
6 components · 5 connections
Explore event streaming platform architecture →
Complex event processing architecture
Specify event time and pattern expiry.
5 components · 4 connections
Explore complex event processing architecture →
Stream enrichment architecture
Define stale-reference behavior explicitly.
6 components · 5 connections
Explore stream enrichment architecture →
Windowed aggregation architecture
Window boundaries change business meaning.
7 components · 5 connections
Explore windowed aggregation architecture →
Event replay architecture
Isolate replay from live side effects.
6 components · 5 connections
Explore event replay architecture →
Message deduplication architecture
Dedup retention must cover redelivery windows.
5 components · 4 connections
Explore message deduplication architecture →
Exactly once processing architecture
State the exact boundary of the guarantee.
6 components · 4 connections
Explore exactly once processing architecture →
Lambda big data architecture
Two processing paths increase reconciliation complexity.
6 components · 6 connections
Explore lambda big data architecture →
Kappa architecture
Retain enough history for reconstruction.
5 components · 4 connections
Explore kappa architecture →
Real time analytics architecture
Trade freshness against computation and storage cost.
5 components · 4 connections
Explore real time analytics architecture →