Your database.
Plus ClickHouse.

Posthouse, Myhouse, and Ighouse add ClickHouse analytics behind your existing relational database — one query layer, no migration. Use the PostgreSQL, MySQL, or Apache Ignite you already run. Offload analytical queries to ClickHouse, reduce load on your primary database, and get columnar performance without changing your application.

Choose Your Stack See How It Works
Three Products, One Architecture

Pick the one that matches your stack

Same engine, same features, same performance. The only difference is which relational database sits on the OLTP side.

PostgreSQL + ClickHouse

Posthouse

For teams running PostgreSQL.

PostgreSQL ClickHouse pgwire Protocol
  • ✓ Connect with psql, any ORM, or JDBC driver
  • ✓ Server-side cursors for streaming exports
  • ✓ PostgreSQL-native types (TIMESTAMPTZ, NUMERIC, TEXT)
  • ✓ Cross-engine joins via coordinator hash-join
Tutorials
MySQL + ClickHouse

Myhouse

For teams running MySQL.

MySQL ClickHouse MySQL Wire Protocol
  • ✓ Connect with mysql CLI, any MySQL connector or ORM
  • ✓ LIMIT/OFFSET pagination for batch exports
  • ✓ MySQL-native types (DATETIME, VARCHAR, DECIMAL)
  • ✓ Cross-engine joins via coordinator hash-join
Tutorials
Apache Ignite 3 + ClickHouse

Ighouse

For teams running Apache Ignite.

Apache Ignite 3 ClickHouse Ignite Thin Client
  • ✓ Native Ignite 3 thin client (MsgPack binary encoding)
  • ✓ JSON-over-TCP wire protocol for any language
  • ✓ Binary tuple encoding with compressed integers
  • ✓ Cross-engine joins via coordinator hash-join
Tutorials
PosthouseMyhouseIghouse
Relational enginePostgreSQLMySQLApache Ignite 3
Analytics engineClickHouseClickHouseClickHouse
Wire protocolpgwire (libpq-compatible)MySQL protocol (mysql-compatible)JSON-over-TCP + Ignite thin client
Outbox throughput~13k rows/s per worker~13k rows/s per worker~96 events/s (1k batch)
Direct push throughput~76k rows/s~70k rows/sComing soon
Streaming APINDJSON, CSV (chunked transfer)NDJSON, CSV (chunked transfer)NDJSON, CSV (chunked transfer)
Client toolposthouse clientmyhouse clientighouse client
ObservabilityPrometheus + /healthz + /v1/status/outboxPrometheus + /healthz + /v1/status/outboxPrometheus + /healthz + /v1/status/outbox
Shared Core

One architecture, three entry points

Posthouse, Myhouse, and Ighouse share the same gateway engine, outbox coordinator, worker, API server, and ClickHouse adapter. The only layer that differs is the relational database driver and wire protocol.

  • ✓ Same hybrid catalog schema
  • ✓ Same outbox event format and worker logic
  • ✓ Same ClickHouse HTTP adapter
  • ✓ Same API endpoints (/v1/query, /v1/export-join, /metrics)
  • ✓ Same SHOW commands (TABLES, EXPORTS, OUTBOX, DEAD LETTERS, SCHEMA DRIFT)
  • ✓ Same output formats (table, JSON, NDJSON, CSV, TSV)
  • ✓ All three pass the same assertion and benchmark suites
Your Application / │ \ pgwire / HTTP mysqlwire / HTTP ignite / HTTP │ │ │ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ │ Posthouse │ │ Myhouse │ │ Ighouse │ │ gateway │ │ gateway │ │ gateway │ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ │ │ │ ┌──────┴──────┐ ┌──────┴──────┐ ┌──────┴──────┐ │ PostgreSQL │ │ MySQL │ │ Ignite │ │ (OLTP) │ │ (OLTP) │ │ (OLTP) │ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ │ │ │ └─────────────────┴─────────────────┘ │ ┌─────────┴─────────┐ │ ClickHouse │ │ (OLAP) │ └───────────────────┘
Features

Everything you need to unify OLTP and OLAP

Stop maintaining two query paths. One connection string for transactional and analytical workloads.

↔

Cross-Engine Joins

Join relational and ClickHouse tables in a single SQL statement. Equality joins with SQL-standard NULL semantics, coordinator hash-join execution.

⚡

Native Wire Protocols

Full pgwire (Posthouse), MySQL wire protocol (Myhouse), and Ignite thin client (Ighouse) support. Connect with your existing tools, ORMs, and drivers — zero code changes.

⇉

Streaming API

Sub-millisecond time-to-first-byte with chunked transfer streaming. JSON, NDJSON, and CSV formats. Buffered mode for small queries, streaming for large exports.

⚙

Reliable Data Transfer

Outbox-pattern replication with at-least-once delivery (~13k rows/s). Direct push mode for bulk loads (~70k rows/s). Retry, dead-letter, and watermark tracking.

🔒

Production Security

Bearer-token API auth, wire protocol password authentication, connection limits, query timeouts, and graceful shutdown with connection draining.

📊

Full Observability

Prometheus metrics for connections, queries, worker throughput, pool state, and process memory. Health and outbox status endpoints built in.

Performance

Built for throughput

Real benchmarks on commodity hardware. Full end-to-end paths including network, serialization, and query execution.

76k
rows/s direct push
10k-row join-export, 130ms P50
13k
rows/s outbox mode
At-least-once delivery, per worker
<1ms
streaming TTFB
Chunked transfer, constant memory
215k
rows/s API query
Buffered JSON, 200k rows, 20 columns
How It Works

Your relational DB is the source of truth.
ClickHouse is your analytics engine.

Posthouse/Myhouse/Ighouse sits between your application and both databases, routing queries to the right engine and keeping data in sync.

1

Register Tables

Declare tables with ENGINE = columnar (ClickHouse-backed) or ENGINE = relational (local DB only). The hybrid catalog tracks everything.

2

Write Normally

Your application writes to PostgreSQL, MySQL, or Ignite as usual. Rows destined for ClickHouse are enqueued in a transactional outbox — same ACID guarantees.

3

Automatic Sync

A background worker claims outbox rows, batches them, and applies to ClickHouse via HTTP. Retry logic and dead-letter handling ensure delivery.

4

Query Either Engine

Run SQL against your relational DB, ClickHouse, or both through a single wire protocol connection or HTTP API. Cross-engine joins are handled transparently.

Tutorials

Learn by doing

Step-by-step guides for every feature. Each tutorial works on Posthouse, Myhouse, and Ighouse with copy-pasteable SQL.

Posthouse

Quickstart

Get running in 2 minutes. Docker Compose up, connect, create a dual-engine table, query it.

 Myhouse

Quickstart

Get running in 2 minutes. Docker Compose up, connect, create a dual-engine table, query it.

 Ighouse

Quickstart

Get running in 2 minutes. Docker Compose up, connect, create a dual-engine table with Ignite 3.

 Posthouse

Client Guide

Architecture, engine modes, catalog introspection, outbox pipeline, join-export, direct push, and reference.

 Myhouse

Client Guide

Architecture, engine modes, catalog introspection, outbox pipeline, join-export, direct push, and reference.

 Ighouse

Client Guide

Architecture, engine modes, catalog introspection, outbox pipeline, join-export with Ignite 3.

 Posthouse

Integrations

Connect with psql, psycopg2, Superset, or any libpq client. Streaming API for large exports.

 Myhouse

Integrations

Connect with mysql CLI, mysql-connector-python, Superset, or PyMySQL. Streaming API for large exports.

 Ighouse

Integrations

Connect with Ignite thin clients, Python, and Superset. Streaming API for large exports.
Pricing

Scale with your data

All plans include Posthouse, Myhouse, and Ighouse. Pricing is based on data flow between engines.

Dev

Build and evaluate

$0
free forever
Data Transfer
  • ✓ 1M rows/mo cross-engine sync
  • ✓ 5M rows/mo API throughput
  • ✓ 1 worker process
Includes
  • ✓ Posthouse + Myhouse + Ighouse
  • ✓ Wire protocols + HTTP API
  • ✓ Cross-engine joins
  • ✓ Community support
Get Started
Coming Soon

Team

Production workloads

$149/mo
billed monthly
Data Transfer
  • ✓ 25M rows/mo cross-engine sync
  • ✓ 100M rows/mo API throughput
  • ✓ Up to 4 worker processes
Includes
  • ✓ Everything in Dev
  • ✓ Prometheus metrics export
  • ✓ Direct push mode
  • ✓ Email support (48hr SLA)
Start Trial
Coming Soon

Business

Growing teams and data

$499/mo
billed monthly
Data Transfer
  • ✓ 250M rows/mo cross-engine sync
  • ✓ 1B rows/mo API throughput
  • ✓ Up to 8 worker processes
Includes
  • ✓ Everything in Team
  • ✓ Graceful shutdown + drain
  • ✓ Query timeout controls
  • ✓ Priority support (24hr SLA)
Start Trial
Coming Soon

Enterprise

Custom scale and support

Custom
annual contract
Data Transfer
  • ✓ Custom sync and throughput limits
  • ✓ Unlimited worker processes
  • ✓ Multi-region deployment support
Includes
  • ✓ Everything in Business
  • ✓ SSO + audit logging
  • ✓ Dedicated support engineer
  • ✓ Custom SLA + onboarding
Contact Sales

Stop maintaining two query paths

PostgreSQL, MySQL, or Apache Ignite — pick the one you run. Add ClickHouse for analytics. One binary, one connection string, two engines.

Start Free Read the Tutorials