Updates to Azure Files: NFS v4.1, higher performance limits, and reserved instance pricing

Azure Files offers fully managed, simple, secure, and serverless enterprise-grade cloud file shares. On the Azure Files team, our mission is to expand Azure Files to more platforms and workloads. We recently took a huge step in workload expansion by announcing the general availability of our NFS v4.1 shares. This greatly expands the workloads you can run on Azure file shares by providing POSIX compatible file systems for Linux virtual machines and container-based workloads. This blog provides information on the general availability of NFS v4.1 shares, increased performance for all premium file shares, and reserved instance pricing for premium file shares to lower your costs.

NFS v4.1 shares are now generally available

In November we announced the general availability of Azure Files support for NFS v4.1. Now you can deploy these fully POSIX compliant, distributed NFS file shares in your production environments for a wide variety of Linux and container-based workloads.

We saw strong interest in the preview with participation from companies of all sizes, ranging from emerging startups to Fortune 100s, running a plethora of workloads. Some examples of workloads include SAP application layer, enterprise messaging, user home directories, custom line-of-business applications, database backups, database replication, AI and machine learning user directories, DevOps pipelines, and many more industry-specific workloads such as the solution from EDF Energy below.

EDF Energy uses Azure file shares as part of its asset management solution:

“At EDF Energy, Nuclear Safety is our overriding priority. As part of our Asset Management solution used to control work, maintenance, and defect resolution to support site license conditions, we needed a performant shared file system between multiple Linux Application Servers—such as an NFS share. We used NFS v4.1 on Azure Files in the preview and have now taken full dependency on it. The NFS system is working very well for us, persisting our files and keeping our IaaS requirements to a minimum.”

—Cathy Handley, AMS Upgrade Programme Manager, EDF Energy—Helping Britain Achieve Net Zero

Customers running critical systems such as SAP have told us that synchronous zonal redundancy is a game-changer for them to achieve high availability for their application layer. With Azure premium file shares you can choose between Locally redundant storage (LRS) or Zonal Redundant Storage (ZRS) redundancy. With ZRS, data is synchronously replicated to three different availability zones within an Azure region. This means your applications’ access to the data will not be disrupted, even in the unlikely event of an entire zone failure. SAP storage administrators gave us great feedback including: “easy to use”, “good performance”, and “better cost optimization.”

Unlike the lower versions of NFS, locking is inbuilt into NFS v4.1. Hence, software like IBM MQ relies on locking support from NFS v4.1 to keep data consistent within a distributed system. While in preview, we enhanced our locking support and implemented locking upgrades and downgrades.

The Azure Files CSI driver (now generally available) makes it easy to access your Azure File shares from Azure Kubernetes Service (AKS). The fast attach-detach times of Azure Files have been appealing to applications that require rapid scale up and scale down.

NFS v4.1 is available in all regions where the premium tier of Azure Files exists. For the full list, see the Azure service availability page. You can now get started using NFS by following these simple step-by-step instructions.

Improved performance

Today, we are announcing more IOPS and throughput for all premium file shares (SMB and NFS).

All shares now provide a minimum of 3000 IOPS, up from the previous 400 IOPS baseline. We are also increasing the minimum burst IOPS such that even the smallest shares can burst up to 10,000 IOPS. Just as before, you will continue to linearly scale IOPS up to 100,000 as the share size increases.

You can now use 100 percent of the provisioned throughput towards either reads or writes. This means you can get up to 10GB/s of read or write traffic. Previously, premium shares used allocated throughput with a 40:60 write:read ratio resulting in max write of 4GB/s and max read of 6GB/s.

These performance enhancements will apply to all existing and new shares across all regions including public and sovereign cloud at no extra cost.

Lower cost with Reserved Instances

All premium file shares (SMB and NFS) now support capacity reservations which provide up to 36 percent discount, by pre-committing to storage utilization.

Reserved instances are also supported for the hot and cool Azure file shares (SMB only).

Get involved

You can put all the updates mentioned in this blog together for NFS v4.1 shares with higher throughput and more IOPS at a lower price. Of course, the performance improvements and reserved instances apply to both NFS and SMB shares. We continue to increase investments in Azure Files and look forward to getting the next wave of updates released.

Source: microsoft.com