Case Study: RedNote accelerates model training and distribution with Alluxio

RedNote accelerates model training by 41% with Alluxio

RedNote, a popular e-commerce and social media platform in Asia, faced significant challenges meeting its nightly 6-hour SLA for updating its machine learning models. Its cloud storage was being throttled, causing training jobs to time out and CPU utilization to drop. The company also needed a faster and more cost-effective method to distribute these large models across its multi-cloud architecture. They implemented Alluxio's distributed cache solution to address these issues.

By deploying a large-scale Alluxio cluster with a 400TB cache, RedNote eliminated its storage bottlenecks. Alluxio accelerated the model training process by 41%, reducing it to 5.5 hours and meeting the SLA. For model distribution, Alluxio provided a 10x speed increase in download times and delivered 80% cost savings compared to their previous solution.

RedNote