Why JuiceFS?
Hundreds of billions of files
In autonomous driving model training, each training involves hundreds of millions of files. JuiceFS' metadata engine is horizontally scalable , which efficiently manages storage for hundreds of billions of files within a single namespace.
Low-latency & high-throughput
For model training, JuiceFS delivers read throughput of hundreds of gigabytes per second, handling hundreds of thousands of files with millisecond-level metadata response time. With flexible cache configurations, JuiceFS provides virtually unlimited aggregate throughput.
Cost Advantage
The volume of data in the autonomous driving sector is increasing rapidly. JuiceFS uses object storage to offer elastic scalability of storage capacity and significantly reduce storage costs. Its flexible architecture minimizes maintenance, and migration expenses.
Feature Overview
A Leading Self-Driving Company Chose JuiceFS over Amazon S3 and Alluxio in the Multi-Cloud Architecture
After thoroughly evaluating factors such as cost, capacity elasticity, and performance, this Self-Driving company chose to construct a storage foundation based on JuiceFS. Currently, the storage platform successfully manages billions of files, achieving an IOPS of 400,000 and a throughput of nearly 40 GB/s. The performance level is comparable to that of the original commercial storage, while significantly reducing storage costs. [Details]
Trusted by Innovators in Autonomous Driving


