by Jacek Migdał
This article explains a complex pricing change announced via email on January 27, 2025. While ClickHouse® has provided an email, blog post, and FAQ, the details can be tricky to follow, so here’s a practical guide to what’s changing.
We estimate a 30% price increase for a typical production workload. This follows a common trend among maturing SaaS products that add more features. Plus, we’re in the post-zero-interest-rate era—VC-subsidized free lunches are over. Customers now have to pay their fair share.
Existing prices have a six-month grandfathering period until July 23, 2025. However, the new pricing introduced egress fees, making migrations from ClickHouse Cloud significantly more expensive. That means now is the time to understand these changes and fully apply all relevant optimizations—this article will help you do just that.
ClickHouse Inc. Business Model
ClickHouse is an open-source analytical database under an Apache License. Like DuckDB, it has a passionate and engaged community that can use it for free indefinitely.
However, ClickHouse isn’t just a hobby project—it raised $250M in Series B funding at a $2B valuation, meaning it needs a solid revenue model. The company follows an open-core approach: the core technology is free, but certain advanced features are locked behind an enterprise proprietary edition, also available as Cloud.
A key architectural advantage of ClickHouse Cloud is compute-storage separation. Initially popularised by data lakes and Snowflake, this approach enables rapid autoscaling and easier maintenance (critical for modern data workloads).
ClickHouse packages all of this into ClickHouse Cloud, a managed service with exclusive features unavailable in the open-source version (or at least very difficult to configure correctly). ClickHouse Cloud seems to be the primary revenue driver for ClickHouse Inc.
Cloud Service Economics
ClickHouse rents servers from major cloud providers like AWS, Azure, and GCP and resells them as a managed service.
The central billing factor is the compute unit, which is priced based on vCPU and GiB of RAM. Essentially, ClickHouse buys computing power in bulk, enhances it with its database service, and then resells it to users at a markup—pretty much like a restaurant buying fruits and serving tasty lemonades.
As an example, let’s break down AWS pricing in us-east-1 as of February 9, 2025.
Links:
In my napkin estimates, I exclude additional costs that ClickHouse incurs to run its cloud—such as monitoring, SRE salaries, and operational overhead. I also ignore any AWS Enterprise Discount Program (EDP) savings, which ClickHouse likely receives for spending $1M+ per year with Amazon.
The storage is mainly sold at no margin. The price is $25.30 / TiB ClickHouse vs. $23 / TiB AWS retail S3 us-east-1. The slight difference in new pricing is because snapshots are now excluded from ClickHouse’s storage pricing, whereas they were previously included. The old price was $35.33/TiB, so this change makes sense, given ClickHouse’s incremental backup system. While the billing logic has shifted, the net cost impact seems neutral.
Previously, ClickHouse didn’t charge for data egress, a rare (and generous) exception since cloud providers charge hefty fees for outbound traffic. With the new pricing, egress fees are now in place—so if you were taking advantage of that loophole, it’s closing.
Most users should not affected as long as you do not export data elsewhere. However, if you dare to migrate away from ClickHouse, be prepared for a hefty “divorce fee” of $115.20 per TiB.
ClickPipes, ClickHouse’s new ETL service, has been enhanced thanks to the PeerDB acquisition. Previously, it was only available in preview, so it makes sense that they’re now charging for it.
On the upside, production-tier scaling is now more granular, and compute-compute separation enables more flexible use cases without requiring extra data copies. This should improve efficiency for many workloads.
Pricing Optimizations
ClickHouse is optimized for analytical workloads (OLAP) but not concurrent transactional workloads (OLTP). For transactions, it is better to use Postgres. As with all tools, use it for what it was designed for.
ClickHouse excels in three key areas:
High ingestion, log processing, and time-series analytics: Great for observability and event-driven workloads. Many observability vendors rely on ClickHouse.
Real-time application dashboards: Ideal for analytical panels in SaaS apps. ClickHouse is uniquely potent here.
Warehouse & data lake analytics: Best performance-to-cost ratio, though join performance and overall platform completeness still trail some industry leaders. However, ClickHouse has been making solid improvements in this area (see Release 24.4 and Release 24.7).
Popular Optimization Strategies
Pick the right primary keys based on query patterns and data cardinality (e.g., by event_time or (toHour(event_time), service_name)). Sorting an order is critical, and changing it later can be painful.
Use materialized views, skip indexes, codecs, and projections to improve performance and cut costs.
Batch ETL jobs smartly to take advantage of autoscaling and avoid unnecessary queries:
ClickHouse Cloud takes time to autoscale (~15 minutes).
Queries running at regular intervals may prevent scaling to zero.
Running all batch jobs within the same hour can reduce compute usage.
Balance sync frequency:
Daily batch syncs are cheaper.
Near real-time Change Data Capture (CDC) provides fresher data but costs more.
Minimize cross-region data transfers by keeping infrastructure in one region. Otherwise, you’ll pay extra directly to the cloud provider or through ClickHouse (see here for details).
Use data retention TTL to cut storage costs. ClickHouse even allows per-column TTL, so you can selectively expire data.
Best Use Cases for ClickHouse Cloud
Even with these pricing changes, ClickHouse Cloud remains one of the most cost-effective data platforms—though it’s less feature-rich and requires some tuning (it's also pretty young). If cost is a concern, it typically outperforms Snowflake, Databricks, and Elastic by an order of magnitude—10x cheaper or more in many cases (read more here).
The Best Migration Patterns
ElasticSearch → ClickHouse: This is my favorite transition. Tools like Quesma provide interoperability, so you can keep using Kibana and existing ingestion integrations while slashing costs by 10x.
Postgres for analytics: Using Postgres for analytical workloads is expensive and inefficient. Historically, moving data out of Postgres was painful and costly (looking at you, Fivetran). But PeerDB (now part of the abovementioned ClickPipes) makes this transition seamless.
Moreover, ClickHouse Cloud has many out-of-the-box optimizations, such as autoscaling or running Graviton, that will pay for themselves. For example, if ClickHouse is used only during work hours, you may pay 20% of the time (40 / (8 * 24)).
So yeah, ClickHouse Cloud just works. Self-hosting is always an option but demands a skilled team and constant tuning. For most teams, the trade-off isn’t worth it.
Supporting Open Source
ClickHouse Inc. is the steward of the open-source version and continues to invest heavily in it. Unlike other companies, ClickHouse uses a permissive Apache license, doesn’t lock previously released features, and even moves some enterprise-only features to open source (e.g., Lightweight Fly Mutations). So, by paying for the cloud, you also support permissive open-source.
Alternatives to ClickHouse Cloud
Hydrolix
The streaming data lake built for high-cardinality, high-dimensional log data. While ClickHouse Cloud is designed for general analytics, Hydrolix is optimized for observability and security workloads, offering faster ingestion and more cost-efficient long-term storage. It also includes Elastic API interoperability via Quesma, making it a strong alternative for users migrating from Elasticsearch. It is also interoperable with Spark and ClickHouse SQL.
Altinity.Cloud
Altinity.Cloud is the fully managed ClickHouse option that remains 100% open source, with no feature restrictions. Unlike ClickHouse Cloud, it supports Bring Your Own Cloud (BYOC) and Bring Your Own Kubernetes (BYOK) deployments, offering managed support while giving users full infrastructure control—including running ClickHouse on Hetzner Cloud for a low-cost alternative. Beyond hosting, Altinity is a major contributor to the ClickHouse ecosystem, maintaining tools like the Kubernetes Operator for ClickHouse to simplify self-hosted deployments.
Edge Delta
Edge Delta is a telemetry pipeline that lets you optimize log ingestion on edge. In particular, you may avoid garbage logs in the first place, reducing your overall cost. You may use it both with ClickHouse Cloud or use Edge Delta observability product based on ClickHouse.
Self-Hosting Open-Source ClickHouse
Running the open-source edition can work if you have:
A primarily steady workload (not too spiky)
Large-scale operations
A highly skilled engineering team
Your data center or a significant cloud discount
I'd be skeptical about this approach unless you have a specialized use case (e.g., air-gapped environments). The operational overhead can be significant.
ByConity
An open-source fork of ClickHouse, initially developed by Alibaba, with a key focus on compute-storage separation—a model popularized by ClickHouse Cloud and Snowflake. ByConity is a good fit for teams needing multi-cloud flexibility while benefiting from ClickHouse’s analytical speed.
BetterStack
An observability platform built on ClickHouse, designed for fast log storage, querying, and monitoring. While ClickHouse Cloud is a general-purpose analytics database, BetterStack is an observability platform with rich user and developer experience.
TinyBird
A real-time analytics layer on top of ClickHouse, allowing developers to turn SQL queries into production-grade APIs. While ClickHouse Cloud is optimized for general analytical workloads, TinyBird specializes in real-time streaming use cases, making it an attractive alternative for SaaS dashboards and interactive applications. Though you will pay extra for that convenience.
Final Thoughts
Overall, ClickHouse Cloud’s pricing is logical and well-thought-out. Seeing more competition in the open-source data space is exciting, as it helps drive innovation and keep costs in check.
If you’re evaluating alternatives, your workload, team expertise, and budget will determine your choice. Feel free to contact us if you have any questions.
ClickHouse is a registered trademark of ClickHouse, Inc.
