ClickHouse Review (2026)

ClickHouse is a column-oriented analytics database that queries billions of rows in under a second. It powers analytics at Cloudflare, Uber, GitLab, and eBay.

What I Like

1. Absurdly fast. Aggregating 1 billion rows in <1 second on a single node. Nothing else comes close for analytical queries.

2. Compression. 10-40x compression ratio. 1TB of raw data becomes 25-100GB stored. This saves significant storage costs.

3. Materialized views. Pre-aggregate data as it arrives. Your dashboard queries hit pre-computed results instead of scanning raw data.

4. SQL interface. ClickHouse SQL is close enough to standard SQL that you can be productive quickly. Some differences, but nothing major.

5. ClickHouse Cloud. Managed service that handles scaling, backups, and maintenance. Generous free tier (10GB).

What I Don't Like

1. Not for OLTP. Don't use ClickHouse as your primary database. No transactions, limited UPDATE/DELETE, not designed for point lookups.

2. JOINs can be tricky. Large JOINs require careful query design. ClickHouse prefers denormalized data.

3. Learning curve. Column-oriented thinking is different. You need to understand merge trees, partitioning, and materialized views.

4. Self-hosting complexity. Running ClickHouse in production requires understanding replication, sharding, and ZooKeeper/ClickHouse Keeper.

When to Use ClickHouse

✅ Product analytics (events, page views, funnels) ✅ Real-time dashboards on large datasets ✅ Log analytics (alternative to Elasticsearch) ✅ A/B test result analysis ✅ Financial data aggregation

❌ Primary application database ❌ Transactional workloads ❌ Small datasets (<1M rows — just use PostgreSQL)

Verdict

9/10 for analytics. Don't use for anything else. ClickHouse is the best open-source analytics database. Period. If you're querying millions of rows for dashboards or analytics, nothing else comes close. Use ClickHouse Cloud for managed, self-host for cost savings at scale. Keep PostgreSQL for your application database.