Accelerate data sharing among serverless containers with Amazon EFS
summary
Shared Amazon EFS provides a regional, serverless POSIX file system that mounts into Lambda, Fargate and Kubernetes, auto‑scales and supports tens of thousands of concurrent clients with gigabytes‑per‑second throughput and 3 GB/s provisioned throughput for heavy cold‑start ML models; IAM enforces mount rights and EFS access points for POSIX identity map containers/functions to specific uid/gid and directories; recent changes claim 5× IOPS, 2× per‑client throughput, 3× read throughput, 3× lower latency; One Zone cuts cost ~47% and IA tiers save up to 92%.
AI-written summary · about
chapters (11)
- 0:00Introduction to EFS with serverless
- 3:07What is Amazon EFS
- 5:14When to use EFS with serverless
- 8:29Content management use cases
- 11:19Machine learning with EFS
- 14:55Data pipelines and file operations
- 16:39Security and identity setup
- 23:21Mounting EFS in containers
- 24:26EFS with Kubernetes CSI
- 25:50Q&A on throughput for ML
- 27:40Best practices and closing
related talks
27mModernize your applications with serverless, containers, and Amazon EFS
31mModernizing your workloads with serverless, containers, and Amazon EFS
09mShared cloud file storage, uses cases, and the latest Amazon EFS innovations
19mWhat's New with Amazon Elastic File System (Amazon EFS)
30mWhat's New in AWS File Storage
18mStorage cost optimization with Amazon EFS
30mWhat's New in File Storage
20mNow Is The Time To Move To AWS Storage