Content Deep Dive

How Infinitas Learning uses ClickHouse Managed Postgres to process billions of events for 1.6M unique learners

Blog post from ClickHouse

Post Details
Company
Date Published
Author
-
Word Count
2,107
Company Posts That Month
20
Language
English
Hacker News Points
-
Post removed?
No
Summary

Infinitas Learning, a major European educational publisher, supports 1.6 million learners who completed 324 million assignments in the 2025–2026 school year, generating billions of events that must be processed and reflected in real time. To unify learner results, the company replaced a complex Kafka-based system with Postgres functions for transactional updates, then adopted ClickHouse Managed Postgres after encountering performance and operational limitations with earlier hosting options; a 750 GB migration took two days with no downtime, and transactional latency fell from milliseconds to double-digit microseconds at the median. For customer-facing analytics, Infinitas paired Postgres with ClickHouse Cloud, which compresses a nearly two-billion-row table from 300 GB to 27 GB, but initially faced unpredictable tail latency caused by cache misses, frequent inserts, and a high-cardinality UUID key. After testing self-hosting and then working with ClickHouse support to tune its data model and settings, the team returned to the managed service with distributed cache, bringing P99 latency to 410 milliseconds—below its 500-millisecond target—while reducing the burden of managing infrastructure and upgrades. Infinitas now uses Postgres for transactions and ClickHouse for analytics, and plans to connect them through the pg_clickhouse extension, with the broader aim of simplifying access to learner data and helping teachers and students benefit from timely insights.

Trends Found in this Post
Trend Post Mentions Total Month Mentions Posts Companies MoM
Real-time 4 1,976 507 170 -53%
Use This Data

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.