Announcing the Public Preview of Metadata Caching for Azure Premium SMB File Shares
Azure Files is excited to announce the public preview of Metadata Caching for the premium SMB file share tier. Metadata Caching is an enhancement aimed at reducing metadata latency for file workloads running on Windows/Linux clients. In addition to lowering metadata latency, workloads will observe a consistent latency experience which will allow metadata intensive workloads to be more predictable and deterministic. Reduced metadata latency also translates to more data IOPS (reads/writes) and throughput.
Once Metadata Caching is enabled, there is no additional cost or operational management overhead when using this feature.
The following Metadata APIs will benefit from Metadata Caching.
- Create: Creating a new file; Up to 30% Faster
- Open: Opening a file; Up to 55% Faster
- Close: Closing a file; Up to 45% Faster
- Delete: Deleting a file; Up to 25% Faster
Workloads that perform a high volume of metadata operations (creating/opening/closing/deleting) against a SMB Premium File share will receive the biggest benefit compared to workloads that are primarily data IO (e.g. databases)
Example of metadata heavy workloads include:
- Web\App Services: Frequently accessed files for CMS\LMS services such as Moodle\WordPress.
- Indexing\Batch Jobs: Large scale processing using Azure Kubernetes or Azure Batch.
- Virtual Desktop Infrastructure: Azure Virtual Desktop\Citrix users with home directories or VDI applications with general purpose file system needs.
- Business Application: Custom line of business or legacy application with “Lift and shift” needs.
- CI\CD - DevOps Pipeline: Building, testing, and deployment workloads such as Jenkins open-source automation.
Expected Performance Improvement with Metadata Cache.
- 2-3x Improved Metadata Latency Consistency
- Improved Metadata Latency beyond 30%
- Increased IOPS and Bandwidth up to 60%
|
How to get started
To begin onboarding to the Public Preview, please sign up on https://aka.ms/PremiumFilesMetadataCachingPreview and additional details will be provided.
Regions Supported
- Australia East
- Brazil Southeast
- France South
- Germany West Central
- Switzerland North
- UAE North
- UAE Central
- US West Central
(Updates will be provided as additional regions are supported and please sign up above to help influence the region prioritization)
Who should Participate?
Whether it is a new workload looking to leverage file shares or existing ones looking for improvements. Any workloads/usage patterns that contains metadata should be encouraged to onboard, specifically metadata heavy workloads that consist primarily of Create/Open/Close or Delete requests.
To determine if your workload contains metadata, can use Azure Monitor to split the transactions by API dimension as described in the following article
Thanks
Azure Files Team
Published on:
Learn moreRelated posts
Announcing the Private Preview of Cosmos DB Azure RBAC Integration
Introduction Managing access to Azure resources often means dealing with two separate permission models: one for management operations and ano...
Azure DocumentDB (with MongoDB compatibility) for Banking: A Modern Customer 360 Approach
Introduction: Transforming Customer Intelligence in Banking Every day, people interact with their bank across mobile apps, branches, call cent...
Exam AI-901: Microsoft Azure AI Fundamentals
With a massive amount of focus on AI across the Microsoft platform, I decided to sit the new AI-901 exam, which is the new Azure fundamentals ...
The problem: All-or-nothing batch processing in Azure Service Bus
Azure Functions lets you settle each Service Bus message on its own within a batch. Complete, abandon, dead-letter, or defer messages one by o...
Welcome to Azure Cosmos DB Conf 2026
Today is the day. Azure Cosmos DB Conf 2026, in partnership with AMD, is a free virtual developer event focused on building modern, scalable a...
Azure Data Studio is retired: Move your Azure SQL workflow to VS Code in 10 minutes
Azure Data Studio retired on February 28, 2026. The recommended path forward is Visual Studio Code with the MSSQL extension. If you used ADS d...