Today, we are introducing EFS Archive, a new storage class for Amazon Elastic File System (Amazon EFS) optimized for long-lived data that is rarely accessed.
With this launch, Amazon EFS supports three Regional storage classes:
- EFS Standard – Powered by SSD storage and designed to deliver submillisecond latency for active data.
- EFS Infrequent Access (EFS IA) – Cost-optimized for data accessed only a few times a quarter, and that doesn’t need the submillisecond latencies of EFS Standard.
- EFS Archive – Cost-optimized for long-lived data accessed a few times a year or less and offering similar performance to EFS IA.
All Regional storage classes deliver gigabytes-per-second throughput and hundreds of thousands of IOPS performance and are designed for eleven nines of durability.
You don’t need to manually pick and choose a storage class for your file systems because EFS lifecycle management can automatically migrate files across storage classes based on their access patterns. This allows you to have a single shared file system that contains files processed in very different ways: from active latency-sensitive to cold rarely-accessed data.
Many datasets have subsets of data that are valuable for generating insights but aren’t often used. With EFS Archive, you can store rarely accessed data cost-effectively while keeping it in the same shared file system as other data. This simplified storage approach allows end users and applications to collaborate on large shared datasets in one place, making it easier and quicker to set up and scale analytics workloads.
Using EFS Archive, you can optimize costs for workloads with large file-based datasets that contain a mix of active and inactive data such as user shares, machine learning (ML) training datasets, SaaS applications, and data retained for regulatory compliance like financial transactions and medical records.
Let’s see how this works in practice.
Using EFS Archive storage
To use the new EFS Archive storage class, I need to configure lifecycle management for the file system. In the Amazon EFS console, I select one of my file systems and choose Edit. To use EFS Archive storage, the file system Throughput mode must be Elastic. Elastic Throughput is the recommended choice for most workloads because it is designed to provide applications with as much throughput as they need with pay-as-you-use pricing.
Now, I configure Lifecycle management to transition files into EFS IA or EFS Archive based on my workload’s access patterns.
My workloads rarely use files older than one month. Files older than a quarter are not used by normal activities but need to be kept for a longer time. Based on these considerations, I select to automatically transition files to EFS IA after 30 days and to EFS Archive after 90 days since the last access. These are the default settings for new file systems.
When one of my old files is accessed, it’s usually an indicator that is being used in a new analysis, so it’ll become active again for some period. For this reason, I use the option to transition files back to Standard storage on their first access in IA or Archive storage.
I save changes, and that’s it! This file system will now automatically use different storage classes based on how files are being processed by my applications.
Things to know
EFS Archive is available today in all AWS Regions where Amazon EFS is offered, excluding those based in China.
To offer a more cost-optimized experience for colder, rarely-accessed files, EFS Archive offers 50 percent lower storage cost than EFS IA with a three times higher request charge when data is accessed. For more information, see Amazon EFS pricing.
You can use EFS Archive with existing file systems by configuring the file system lifecycle policies. New file systems are created by default with a lifecycle policy that automatically transitions files to EFS IA after 30 days and to EFS Archive after 90 days since the last access.
Optimize your storage costs by configuring lifecycle management for your Amazon EFS file systems.
— Danilo
