StreamHouse
Phase 7 Released — Exactly-once semantics

Event streaming
for the cloud era

Legacy streaming infrastructure is too heavy, slow, and complicated. StreamHouse is S3-native, serverless, and ready.

StreamHouse

Sources

Kafka
Postgres
AWS

Destinations

ClickHouse
Spark
Flink

Deployment Options

Self Hosted
BYOC
Cloud
The Problem

Kafka wasn't built for the cloud

Traditional streaming requires expensive disks, complex replication, and constant operational overhead.

3x storage cost for replication
ZooKeeper/KRaft dependencies
Manual capacity planning
Expensive cross-AZ traffic
The Solution

S3-native from day one

StreamHouse uses object storage as the source of truth. Stateless agents. Zero replication. Infinite scale.

11 nines durability via S3
80% lower infrastructure costs
Instant, automatic scaling
No disks to manage
Architecture

Disaggregated by design

Compute and storage scale independently. Agents are stateless. S3 provides durability.

Stateless Agents

Ephemeral compute that can be killed or restarted without data loss. No persistent state on the broker.

S3 Storage

Events written directly to object storage with LZ4 compression. 11 nines durability, infinite capacity.

Metadata Store

Lightweight coordination via PostgreSQL. Only offsets, leases, and watermarks—no event data.

Features

Everything you need

S3-Native Storage

Direct writes to object storage with LZ4 compression.

Zero Replication

S3 handles durability. No inter-broker traffic costs.

SQL Processing

Built-in stream processing. No separate Flink cluster.

Stateless Agents

Kill, restart, or scale freely. All state in S3.

Smart Caching

LRU cache reduces S3 reads by 80-95%.

Full Observability

Prometheus metrics and Grafana dashboards.

Developer Experience

Up and running in minutes

Start streaming with a few commands. No complex configuration, no 50-page runbooks.

terminal
# Start infrastructure
$ docker compose up -d
✓ Services ready
# Launch agent
$ cargo run --bin agent
✓ Listening on :9090
# Produce events
$ streamctl produce events -v '{"id":1}'
→ offset=0
Use Cases

Built for modern data teams

Analytics

Real-time analytics

Replace Kafka → Flink → ClickHouse with integrated SQL processing.

Learn more
Observability

Log aggregation

Store logs in S3 at $0.023/GB. Infinite retention.

Learn more
ML/AI

ML feature pipelines

Compute real-time features with SQL. Direct S3 output.

Learn more
Data Sync

Change data capture

Simplified CDC with built-in transformations.

Learn more

Ready to simplify
your streaming?

Join teams cutting infrastructure costs by 80% while eliminating operational complexity.