Home / Companies / OpenObserve / Blog / Post Details
Content Deep Dive

How Yupao Moved 69 TB of Logs a Day Off Alibaba Cloud SLS to Self-Hosted OpenObserve

Blog post from OpenObserve

Post Details
Company
Date Published
Author
OpenObserve
Word Count
2,400
Company Posts That Month
10
Language
English
Hacker News Points
-
Post removed?
No
Summary

Yupao Technology, operator of the Yupao Zhipin recruitment platform, migrated much of its logging workload from Alibaba Cloud SLS to a self-hosted OpenObserve deployment after annual SLS costs exceeded $140,000 and logs were fragmented across separate VPC projects. Over five overlapping phases lasting five months, the company used dual-writing, developer testing, a query-mapping handbook, and gradual production rollout to reduce migration risk for more than 100 engineers. Its architecture uses Vector for collection and processing, Redpanda as a shared cross-VPC ingestion buffer, Kubernetes-hosted OpenObserve components, and Alibaba Cloud OSS for Parquet-based storage; it reportedly ingests 69 TB of raw logs daily, retaining 130.7 billion log lines with an 8.1:1 compression ratio. Yupao estimates that the self-hosted stack costs about $40,000 annually, saving more than $60,000 in its first year while providing a unified log view, configurable retention and indexing, and isolated alert-query capacity. The company retained selected logs in SLS, including error logs tied to an established SMS and phone alerting chain, gateway logs that can only be sent to SLS, and some streams not yet migrated, reflecting its stated priority of balancing lower costs with operational stability.

Trends Found in this Post
Trend Post Mentions Total Month Mentions Posts Companies MoM
Observability 12 2,762 731 188 -15%
Kubernetes 8 4,363 405 103 +23%
OpenTelemetry 4 708 130 59 -11%
LLM 2 6,033 1,166 244 +15%
Real-time 1 4,852 1,086 241 +4%
Secrets Management 1 2,046 504 139 -11%
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.