Tinybird vs. ClickHouse

ClickHouse is an open source columnar database ideally suited for real-time analytics. It's fast, powerful, and flexible for a wide range of analytics use cases.

Tinybird is a data platform for user-facing analytics built on top of open source ClickHouse. It offers the performance and functionality of the ClickHouse database in a convenient abstraction that reduces developer friction and shortens time to production.

We're often asked how Tinybird is different from ClickHouse, and when developers should choose Tinybird over ClickHouse (open source or managed), and vice versa.

Here's the TL;DR:

You choose Tinybird if:

• You want a fast and scalable analytics backend
• You want a highly-tuned ClickHouse instance ready out of the box
• You don't want to build and manage infrastructure
• You don't care to learn deep ClickHouse internals
• You want to focus on building a use case in production
• You need to ship quickly

You choose ClickHouse if:

• You want a fast and scalable DBMS
• You want more control over how to scale and tune your cluster
• You want to build and control your own analytics service
• You want to deeply learn ClickHouse
• You can take the time to understand ClickHouse internals
• You want to host your database on-premises

If you're ready to try Tinybird based on that info, you can sign up here. There's a generous free tier with no time limit and no credit card required.

If you want more detail, you can keep reading.

What is ClickHouse?

ClickHouse is an open source, column-oriented SQL database management system. It was originally developed by Alexey Milovidov during his time at Yandex to power Yandex.Metrica, the second largest web analytics platform in the world. In 2016, ClickHouse was open sourced by Yandex and has been used for real-time analytics at scale by companies like Uber, eBay, Cisco, Spotify, Netflix, and many others.

What is Tinybird?

Tinybird is a serverless real-time data platform for building user-facing analytics. It was founded in 2019 with a mission to give developers tools for working with real-time data at scale. Tinybird uses ClickHouse as its underlying database technology, providing abstractions and services on top of the database to make it easier to use. Our company includes many contributors to the ClickHouse codebase, and we frequently ship important improvements to the open source project. Engineers at companies like FanDuel, Canva, and Vercel use Tinybird to build and ship user-facing analytics features in their respective apps.

How to compare Tinybird and ClickHouse

Let's start by framing the discussion so it's clear how we're comparing Tinybird and ClickHouse. Here are the factors we'll consider:

1. Infrastructure & Database Management
2. Performance
3. Cost
4. Developer Experience

We'll assess Tinybird vs. ClickHouse along these four axes in an attempt to give a complete and accurate comparison.

Infrastructure & Database Management

As with many open source database management systems, ClickHouse can be self-hosted (on-prem or cloud) or purchased as a managed service through a vendor.

I won't belabor the pros and cons of self-hosting a database or buying a managed version. If you are dead set on self-hosting your database, then Tinybird is not for you. Just work with the open source version of ClickHouse.

If you intend to use a managed ClickHouse service, or are at least considering it, then you have options. For the sake of this discussion, we'll focus on two of the most popular: Tinybird and ClickHouse Cloud.

Similarities in how Tinybird and ClickHouse manage infrastructure

Both Tinybird and ClickHouse Cloud offer a managed ClickHouse service that simplifies the job of maintaining a database.

Both are available on multiple clouds (AWS, GCP, Azure) across various global regions.

Both handle automatic updates on the database technology as new features are merged and released.

Both offer a web-based UI for interacting with the database with varying levels of abstraction.

Both provide some level of services on the write and read sides of the database, including various managed ETLs, static endpoints for queries, etc.

In addition, Tinybird and ClickHouse Cloud share most of the underlying features of the open source ClickHouse database, such as the SQL language and materialized views.

Differences in how Tinybird and ClickHouse manage infrastructure

In simple terms, Tinybird is a higher level of abstraction than ClickHouse Cloud. Tinybird offers an out-of-the-box tuned database cluster optimized for user-facing analytics use cases. Many of the finer details of database management are abstracted and not exposed to the end user.

ClickHouse Cloud exposes more database internals, offers BYOC in preview

ClickHouse Cloud gives the user more control over the underlying database settings. With ClickHouse Cloud, you can connect directly to the ClickHouse client, and you have more control over how you tune your database for your use case.

ClickHouse Cloud has also recently opened a "Bring Your Own Cloud (BYOC)" waitlist for private preview on AWS, which Tinybird currently does not offer.

While Tinybird exposes most of the underlying database features, a few niche features are purposefully disabled for performance or security reasons. For example, Tinybird does not support a small set of SQL functions, a few data types, and some of the ClickHouse Table Engines. For 99% of use cases, this is not problematic, but it bears mentioning.

Tinybird carefully manages database upgrades

Tinybird also handles a lot of the complexities of upgrading the database. As with any software, updates can introduce regressions or breaking changes. Whenever a ClickHouse update is released, Tinybird typically upgrades clusters within a few days. 

Tinybird checks all existing customer queries and table schemas against the main branch daily to check for regressions and prevent breaking changes. The upgrade process is done in phases so that only parts of a cluster are changed, and read and write paths are checked along the way to ensure compatibility and high availability.

Tinybird offers more built-in services for ingestion, integration, API hosting, and observability

Beyond database management, Tinybird offers more ancillary features alongside the database so you don't have to build them yourself.

On the ingestion side, Tinybird has more native data source connectors, includes a managed HTTP streaming endpoint, and provides built-in infrastructure for queuing and handling on-demand and scheduled import and export jobs.

In addition, Tinybird provides more services on the read side of the database. Tinybird is famous for its "single-click" API generation flow, in which you can instantly publish any query as a static, named, and secured API endpoint complete with dynamic query parameters, auto-generated OpenAPI 3.0 specs, and support for both static access tokens and JWTs to secure read access with row-level security policies.

Tinybird also offers built-in observability at both the database level and the UI. Every job, call, or query is logged in a Service Data Source, which can be queried free of charge. Tinybird users can build APIs to monitor ingestion throughput, identify API errors, alert on failed jobs, and more. Tinybird offers a built-in time series charting UI for log analysis and integrates well with observability tools like Grafana.

In effect, Tinybird minimizes your development surface area. If you choose Tinybird, you will need to build or source fewer additional services than you otherwise would with self-hosted or managed ClickHouse.

A concise comparison of infrastructure features between Tinybird and ClickHouse Cloud is included below for reference:

Performance

Tinybird is built on top of the Clickhouse open source database, and so the performance you get from Tinybird will be on par with what you would get from any hosted ClickHouse instance running on similar compute. ClickHouse's performance benchmarks readily apply here.

Performance does depend on compute, of course, so the nature of how ClickHouse is hosted will impact performance. Tinybird doesn't eschew performance for ease of use. In fact, Tinybird's engineers regularly contribute performance improvements to the open source ClickHouse project (examples here, here, and here).

With Tinybird, all non-Enterprise plans are hosted on shared infrastructure that is scaled to meet demand at a high level of performance for the vast majority of use cases. Even on shared infrastructure, many Tinybird customers readily achieve 20+ MB/s write throughput and 1K+ QPS with <50ms p95 query latency out of the box.

Where more resources are needed, Tinybird's Enterprise plans allow the option for dedicated infrastructure that can be independently scaled, further extending Tinybird's performance to the limits of ClickHouse itself.

And though Tinybird can scale compute as needed to handle the load, we also focus heavily on optimization. We have a team of data engineers that work directly with clients to continually optimize queries to increase compute efficiency without adding more cores. For our clients, that means faster queries without consuming more infrastructure credits or usage quotas.

Other managed versions of ClickHouse, such as ClickHouse Cloud, are designed to give you more control over compute size, and are priced according to the size of compute you select. You can choose the machines you need to provide the necessary performance for your use case. Tinybird chooses to abstract this selection and handle scaling compute on an as-needed basis.

Cost

The cost of managing ClickHouse can be expressed in terms of infrastructure (storage and compute) and development time for setting up, maintaining, and scaling the database.

Cost of self-hosting ClickHouse

Self-hosting ClickHouse will (hopefully) result in lower infrastructure costs but at the expense of much higher development costs to manage the cluster.

If you want to better understand what it costs to self-host ClickHouse, check out this resource from the GitLab Handbook, which includes GitLab's assessment of costs associated with their own self-hosted ClickHouse instance. There's a full GitLab issue with cost tables and explanations of all the variables that influence self-hosting costs.

If reading that GitLab issue feels as cumbersome to you as it does to me, then you are probably leaning toward a managed service that makes understanding costs much more predictable while also removing at least some of the associated development costs.

Cost of ClickHouse Cloud

Managed versions of ClickHouse such as ClickHouse Cloud are designed to abstract some, but not all, of the complexities of managing the underlying database. With ClickHouse Cloud, you'll pay for the cost of storage and compute plus any additional time spent fine-tuning the database or building ancillary services.

ClickHouse Cloud does not offer a free plan, but development clusters clamped at 16 GiB RAM, 2 vCPUs, and 10 GB of storage will cost up to $160 per month depending on cloud provider and active hours. A production-grade cluster with 96 GiB RAM, 24 vCPUs, and 500 GB of storage will cost up to $2,500 per month depending on cloud provider and active hours. You can calculate ClickHouse Cloud pricing based on storage and compute size with the ClickHouse pricing calculator.

Cost of Tinybird

Tinybird is designed to eliminate the development costs associated with managing a database while offering comparably priced access to storage and compute.

Tinybird offers both zero-to-scale usage-based pricing and infrastructure-based pricing. All non-Enterprise plans are subject to standard usage-based pricing at $0.34/GB of compressed storage and $0.07GB of processed data.

Additionally, Tinybird offers a free development plan that includes 10 GB of storage, unlimited data processing, and up to 1,000 API requests per day on published Endpoints. 

93% of non-enterprise Tinybird customers on paid plans pay less than $100/month for their production Workspaces. The median non-Enterprise paid Tinybird Workspace costs less than $10/month.

Tinybird's Enterprise customers can choose to have access to dedicated infrastructure and can select pricing plans based on usage quotas or infrastructure credits. All Enterprise plans come with steep volume discounts for storage and compute. 

For more information about Tinybird pricing, check out the public pricing page or find more detailed explanations of how usage is calculated in the pricing docs.

Developer Experience

If you could capture the Tinybird ethos in a sentence, it's this: "Fast databases make queries fast. Tinybird makes developers fast." Tinybird intentionally abstracts the complexities of both ClickHouse and the additional infrastructure you'd have to build for a fully-baked analytics service so that you can become productive more quickly.

Put simply, developers choose Tinybird over ClickHouse because they want to focus on their use case and shorten their time to production. They want the performance that ClickHouse offers, without the effort to manage it.

Tinybird offers time-saving features that self-hosted and managed ClickHouse versions don't, including:

• Single-click API generation, API hosting, API documentation, token management, and advanced templating and logic that extend ClickHouse SQL
• A managed HTTP streaming endpoint that supports up to 1,000 requests per second out of the box.
• A managed Kafka consumer that scales horizontally and handles streaming writes in real time.
• Managed data warehouse connectors (BigQuery, Snowflake) that handle authentication and scheduled import job queues.
• Managed job queues for exports to Amazon S3 for long-term storage.
• Managed job queues for copying the results of queries into database tables, using both replacing and append strategies
• A CLI, plain text file structure, and native git integration for managing Tinybird resources in production as a part of git-based, CI/CD workflows.

A summary of time-saving Tinybird features is included below for reference.

Choosing Tinybird vs. ClickHouse

Tinybird, managed ClickHouse, and self-hosted ClickHouse all provide excellent performance for real-time analytics use cases.

The choice between these options will depend on your goals and constraints: desired speed to market, budget, time, and engineering resources. 

Those with time, budget, and solid data engineers with a desire to learn ClickHouse might choose to self-host. 

Those who want a more classic ClickHouse experience on hosted infrastructure might choose a managed ClickHouse like ClickHouse Cloud.

Those who want to harness the performance of ClickHouse to quickly ship user-facing analytics features in production might choose Tinybird.

Need more help deciding? Feel free to request a Tinybird demo or just sign up for a free Tinybird plan to give it a spin.