应用场景

When we moved into the LLM observability and analytics space, we decided to back LangSmith with ClickHouse instead of Postgres. Ankush, CTO of LangChain

ClickHouse helps us efficiently and reliably analyze logs across trillions of Internet requests to identify malicious traffic and provide customers with rich analytics.

Now, our customers can search through months of browser and server-side log data in under a second thanks to the tech behind ClickHouse.

ClickHouse was perfect as Big Data Storage for our ML models

We store approximately 3 billion events (rows) per day at a rate of approximately 2 million events per minute. We also have to serve some pretty complex data visualizations that depend heavily on filtering very large amounts of data and calculating complex aggregations in a reasonably fast time frame for the sake of the user experience.

At Sony Entertainment Television, we ingest tens of millions of CDN records into ClickHouse Cloud and run millions of queries against them daily. This allows our operations team to monitor the delivery of our content in real-time, and analyze/investigate potential issues the moment they arise. ClickHouse Cloud has helped us to optimize costs and ensure the high availability and resilience of our services.

QRadar Log Insights uses a modern open-source OLAP data warehouse, ClickHouse, which ingests, automatically indexes, searches and analyzes large datasets at sub-second speed. You get near real-time visibility and insights from your ingested data.

The platform is ingesting millions of logs per second from thousands of services across regions, storing several PBs worth, and serving hundreds of queries per second from both dashboards and programs.

ClickHouse is a fast and highly performant analytical database, widely used across Instacart to power other use-cases such as critical retailer and ads dashboards, calculating results for A/B testing, and machine learning signals.

We evaluated more than a dozen different big data systems before settling on ClickHouse.

ClickHouse’s performance exceeds all other column-oriented database management systems.

We collect tens of thousands of data points from customers’ phones and other more traditional sources. ClickHouse is used as a way to process all of these SMS messages and extract valuable information used for the scoring and fraud models.

Moving from Elasticsearch to ClickHouse was a long journey, but this is one of the best tech decisions we ever took.

At Lyft, we ingest tens of millions of rows and execute millions of read queries in ClickHouse daily with volume continuing to increase. On a monthly basis, this means reading and writing more than 25TB of data.

Cognitiv uses ClickHouse to power their ML offline feature store for its blazing speed and resource efficiency.

“Training these models requires immense computational power and the ability to handle and analyze vast quantities of data quickly and efficiently.

As a smaller company with large datasets, cost is important to us,” explains Jason from Cognitiv. “ClickHouse is fast, but its real value is in letting us better utilize our resources. Basically, we don’t need to spend as much money to solve the same problem.”

“By utilizing expert models and embeddings, we detect substantive changes in web pages and identify connections between pages that share similar characteristics.”

ClickHouse has proved to be a game-changer, propelling us towards greater efficiency and effectiveness in managing our data infrastructure.

Ved Surtani, VP, Engineering, Platform & Architecture at Tekion

We had a billion rows to store...

Adopting ClickHouse has enhanced our data analytics capabilities, supporting the growing demands of our internal teams efficiently and cost-effectively. Frank Chen, Expert OLAP Engineer at Shopee

Switching to a Kafka and ClickHouse-based architecture simplified our operations and reduced costs by enhancing performance and enabling real-time, large-scale data processing

We have saved costs, savings not to be sniffed at, but that was not the driving factor. This was a qualitative step. We just could not do the things we wanted until we had ClickHouse and that is why we’re so excited about it.

In the post-evaluation of each database against our criteria (with metrics ranging from query performance to cost), ClickHouse emerged as the unrivaled frontrunner. It excelled across the board, even astonishingly so in certain domains, and proved more cost-efficient.

With ClickHouse, the data pipeline logic is simplified, and is only dealing with the “streaming” aspect of the write as opposed to all of these complexities. ClickHouse thus enables a simpler write design pattern just like any other new age data lake systems like Hudi etc. but with a more simplistic developer experience.

At DeepL, we use ClickHouse as our central data warehouse.

Best technical decision we ever made