Uber: Serving Real-Time App Crash Analytics While Saving $2M+ With Apache Pinot
StarTree News, Thursday, January 18th, 2024
For a company like Uber, real-time data is the lifeblood of both their customer-facing and internal services. Customers rely on real-time data to get rides and grab food at their convenience. Internal teams also depend on up-to-date data to power the infrastructure behind their customer-facing applications, such as for their in-house tool that monitors mobile app crash analytics.
Uber migrated to Apache Pinot to power this in-house tool and experienced significant improvements compared to their previous analytics engine (Elasticsearch). By moving to Pinot, a true real-time analytics platform, Uber saw benefits including:
- 70% reduction in infrastructure costs (a savings of $2M+ yearly)
- 80% reduction in CPU cores
- 66% reduction in data footprint
- 64% reduction in page load times (from 14 seconds to under 5 seconds)
- Reduced ingestion lag to <10 milliseconds
- Decline in query timeouts and elimination of data loss issue
Watch now
The content of this blog is based on an in-person meetup that featured Apache Pinot user stories. We also reference a blog from Uber's engineering team that covers how they use Pinot to serve real-time analytics for mobile app crashes.