Introducing ClickHouse® on Exoscale: Real-Time Analytics
Clovis Genevard
ClickHouse® is now part of the Exoscale DBaaS catalog as a managed service.
Our new product ClickHouse® is an open-source columnar database built for real-time analytical queries at scale. We have partnered with Aiven to bring it to our European zones as a fully managed service, so you can run fast analytics on massive datasets without managing the infrastructure yourself.
New services run ClickHouse® version 25.8 LTS, a release that matters: vector search is now generally available, projections got lighter, and SQL compatibility took a real step forward. Here is what you get.
Quick summary of this article
- ClickHouse® is now part of the Exoscale DBaaS catalog as a managed service, in partnership with Aiven, available across Exoscale European zones.
- ClickHouse is an open-source columnar database built for real-time analytical queries at scale. Sub-second aggregations on billions of rows, without managing the infrastructure yourself.
- Services run ClickHouse 25.8 LTS, which brings generally available vector search with the HNSW index, lightweight projections, correlated subqueries, and lightweight updates in beta.
- Tiered storage moves older data to Exoscale Object Storage in the same zone, and federated queries read from Managed PostgreSQL and object storage directly in SQL.
- A 99.99% uptime SLA is available on Business and Premium plans. During early access you create services with the Exoscale CLI or API, with free trial credit for new accounts.
Why ClickHouse on Exoscale
The challenge is not collecting data. It is querying it fast enough to act on it. Whether you are building real-time dashboards, monitoring IoT telemetry, or feeding AI features, you need a database that handles billions of rows efficiently.
Managed ClickHouse® on Exoscale gives you:
- Sub-second query performance: Interactive SQL on massive datasets, fast enough for real-time dashboards and user-facing applications
- Data that stays in Europe: All data remains in your chosen European zone, supporting data residency and GDPR requirements
- Familiar interfaces: Connect over native TCP, HTTPS, the MySQL protocol or Arrow Flight, with client libraries for Python, Go, Node.js and more
- Predictable storage costs: Tiered storage automatically moves data from fast local disk to S3-compatible Exoscale Object Storage in the same zone as local storage fills, with optional per-table TTL policies
- Federated queries: Read from Managed PostgreSQL with ClickHouse’s PostgreSQL table engine, and query S3-compatible object storage directly in SQL
The product page covers the full feature set and plans in detail.
What’s in ClickHouse 25.8 LTS
ClickHouse 25.8 LTS brings new capabilities for faster analytics, broader SQL compatibility, and modern AI workloads.
Vector search, now generally available
Approximate nearest-neighbor search using the HNSW index is now GA. Binary quantization for the vector similarity index drastically reduces memory consumption and improves indexing speed for large-scale embedding workloads: RAG pipelines, recommendation systems, and semantic search.
If you already store event or product data in ClickHouse, you can now serve vector similarity queries from the same engine, without adding a separate vector database to your stack.
Lightweight projections
ClickHouse has historically been optimized around a single primary index per table. Lightweight projections now work as secondary indexes: they store only the sorting key and a pointer back to the base table, rather than duplicating the full dataset. Queries that filter on columns outside your primary key get faster without the storage overhead of full materialized projections.
Broader SQL compatibility
Correlated subqueries are now supported, alongside lag and lead window functions. This reduces rewrite effort when migrating analytical workloads from PostgreSQL, BigQuery, Snowflake, and other SQL-based systems.
Lightweight updates, currently in beta
ClickHouse 25.8 introduces UPDATE ... SET ... WHERE directly on MergeTree tables using a patch parts mechanism, instead of the classic mutations that rewrite entire columns. Single-row and small-batch corrections become practical: correcting records, updating statuses, or maintaining slowly changing dimensions. The feature is in beta, so test it against your workload before relying on it in production.
What you can build
Observability. Ingest high-volume log and metric events from OpenTelemetry or your existing event pipeline, then query them in real time without placing analytical load on your application database.
Real-time reporting. Power user-facing dashboards with sub-second queries, even across very large datasets.
AI pipelines. Aggregate high-volume data streams into feature pipelines and use vector search to support responsive AI applications as your data grows.
Powered by Aiven’s managed open-source expertise
Exoscale’s DBaaS is powered by Aiven, one of Europe’s leading companies for managing open-source data infrastructure in the cloud. The partnership brings Exoscale customers a managed ClickHouse® experience backed by Aiven’s operational expertise, within Exoscale’s European cloud environment.
A 99.99% uptime SLA is available on Business and Premium plans.
Get started
The service is in early access across Exoscale zones in Europe, with per-second billing and no minimum term. You create and manage services with the Exoscale CLI or API today; Portal integration and the dedicated exo dbaas commands arrive with general availability. Compare plans on the pricing page.
ClickHouse® is a registered trademark of ClickHouse, Inc. https://clickhouse.com. Exoscale is not affiliated with, sponsored by, or endorsed by ClickHouse, Inc.
