Merge pull request #7836 from MicrosoftDocs/copilot/clarify-csv-file-system-support

Clarify CSV file system guidance: NTFS for SAN, ReFS for S2D

Author: prmerger-automator[bot]

WindowsServerDocs/failover-clustering/clustering-requirements.md

@@ -61,13 +61,16 @@ In most cases, attached storage should contain multiple, separate disks configur
 
 Some clusters have a disk that functions as the *disk witness*, which is a disk in cluster storage that holds the cluster configuration database. Failover clusters only have disk witnesses if the disk witness is supposed to be part of the specified quorum configuration. For more information, see [Understanding Quorum in Storage Spaces Direct](/azure/azure-local/concepts/quorum?context=/windows-server/context/windows-server-storage). The other disks contain required files for clustered roles.
 
-You should also consider these additional storage requirements:
+You should also consider these other storage requirements:
 
 - When you use native disk support included in failover clustering, you should only use basic disks, not dynamic disks.
-- We recommend you format disk partitions with New Technology File System (NTFS). If you use Cluster Shared Volumes (CSV), you must format the partitions for each CSV with NTFS.
+- We recommend you format disk partitions with New Technology File System (NTFS). If you use Cluster Shared Volumes (CSV), you must format the partitions for each CSV with NTFS or Resilient File System (ReFS) based on the following guidelines:
+
+  - **SAN volumes**: Format with NTFS before adding to CSV. NTFS enables Direct I/O mode for better performance.
+  - **Storage Spaces Direct (S2D) volumes**: Format with ReFS before adding to CSV. ReFS provides data integrity features and optimizations for Storage Spaces Direct.
 
   >[!NOTE]
-  >If you have a disk witness for your quorum configuration, you can format the disk with either NTFS or Resilient File System (ReFS).
+  >If you have a disk witness for your quorum configuration, you can format the disk with either NTFS or ReFS.
 
 - For the partition style of the disk, you can use either a Master Boot Record (MBR) or a GUID Partition Table (GPT).
 

WindowsServerDocs/failover-clustering/failover-cluster-csvs.md

@@ -15,16 +15,6 @@ CSV offers a clustered file system that operates on top of NTFS or ReFS. Typical
 
 - Clustered virtual hard disk (VHD/VHDX) files for a clustered Hyper-V virtual machine (VM).
 
-- Scale-out file shares to store application data for the Scale-Out File Server clustered role. Examples of the application data for this role include Hyper-V VM files and Microsoft SQL Server data. ReFS isn't supported for a Scale-Out File Server in Windows Server 2012 R2 and previous releases. For more information about Scale-Out File Server, see [Scale-Out File Server for Application Data](sofs-overview.md).
-
-- Microsoft SQL Server 2014 (or higher) Failover Cluster Instance (FCI). CSVs don't support the Microsoft SQL Server clustered workload in SQL Server 2012 and earlier versions of SQL Server.
-
-- Windows Server 2019 or higher Microsoft Distributed Transaction Control (MSDTC).
-
-In later releases of Windows Server, CSV functionality is enhanced. For example, dependencies on Active Directory Domain Services (AD DS) were removed. Support was added for the functional improvements in `chkdsk` for interoperability with antivirus and backup applications, and for integration with general storage features such as BitLocker-encrypted volumes and Storage Spaces. For an overview of CSV functionality that was introduced in Windows Server 2012, see [What's New in Failover Clustering in Windows Server 2012](/previous-versions/windows/it-pro/windows-server-2012-r2-and-2012/dn265972(v%3dws.11)).
-
-Windows Server 2012 R2 introduced additional functionality, such as:
-
 - Distributed CSV ownership with improved diagnosability
 
 - Increased resiliency through availability of the Server service
@@ -33,6 +23,14 @@ Windows Server 2012 R2 introduced additional functionality, such as:
 
 - Enhanced interoperability that includes support for ReFS and deduplication.
 
+- Scale-out file shares to store application data for the Scale-Out File Server clustered role. Examples of the application data for this role include Hyper-V VM files and Microsoft SQL Server data. ReFS isn't supported for a Scale-Out File Server in Windows Server 2012 R2 and previous releases. For more information about Scale-Out File Server, see [Scale-Out File Server for Application Data](sofs-overview.md).
+
+- Microsoft SQL Server 2014 (or higher) Failover Cluster Instance (FCI). CSVs don't support the Microsoft SQL Server clustered workload in SQL Server 2012 and earlier versions of SQL Server.
+
+- Windows Server 2019 or higher Microsoft Distributed Transaction Control (MSDTC).
+
+To learn more about CSV functionality introduced in previous versions, see [What's New in Failover Clustering in Windows Server 2012](/previous-versions/windows/it-pro/windows-server-2012-r2-and-2012/dn265972(v%3dws.11)).
+
 For information about using data deduplication on CSVs for Virtual Desktop Infrastructure (VDI) scenarios, see blog posts [Deploying Data Deduplication for VDI storage in Windows Server 2012 R2](https://techcommunity.microsoft.com/t5/storage-at-microsoft/deploying-data-deduplication-for-vdi-storage-in-windows-server/ba-p/424777) and [Extending Data Deduplication to new workloads in Windows Server 2012 R2](https://techcommunity.microsoft.com/t5/storage-at-microsoft/extending-data-deduplication-to-new-workloads-in-windows-server/ba-p/424787).
 
 ## Requirements and considerations for using CSV in a failover cluster
@@ -51,6 +49,11 @@ Before using a CSV in a failover cluster, review the network, storage, and other
 
   In Windows Server 2012, a disk or storage space for a CSV must be a basic disk that is partitioned with NTFS. In Windows Server 2012 R2, a disk or storage space for a CSV must be a basic disk that is partitioned with NTFS or ReFS. In Windows Server 2016 and later, and Azure Local, a disk or storage space for a CSV must be either a basic disk or GUID Partition Table (GPT) disk that is partitioned with NTFS or ReFS. A best practice is to also review the guidelines that are provided by your storage vendor.
 
+  For optimal performance and functionality, use the following file system guidelines when formatting disks before adding them to CSV:
+
+  - **SAN volumes**: Format with NTFS before adding to CSV. NTFS enables Direct I/O mode for SAN-attached storage, which provides better performance. CSVs formatted with ReFS on SANs operate in redirected I/O mode.
+  - **Storage Spaces Direct (S2D) volumes**: Format with ReFS before adding to CSV. ReFS provides data integrity features, block cloning, and optimizations designed for Storage Spaces Direct workloads.
+
   - In Windows Server 2012, you can't use a disk as a CSV formatted as FAT, FAT32, or ReFS.
 
   - In Windows Server 2012 R2 and later, you can't use a disk as a CSV formatted as FAT or FAT32.

WindowsServerDocs/failover-clustering/failover-cluster-manage-cluster-shared-volumes.md

@@ -21,7 +21,7 @@ To manage CSVs, ensure the following prerequisites are met:
 
 - Your cluster configuration must have at least two nodes configured in a failover cluster. All nodes must be joined to the same Active Directory (AD) domain.
 
-- At least one shared disk (LUN) must be accessible by all cluster nodes and configured as a cluster disk using NTFS or ReFS. If you're using iSCSI or Fibre Channel for shared storage, ensure that multipath I/O (MPIO) is properly configured to provide redundancy and optimal performance.
+- At least one shared disk (LUN) must be accessible by all cluster nodes and configured as a cluster disk using NTFS or ReFS. Format SAN volumes with NTFS and Storage Spaces Direct (S2D) volumes with ReFS for optimal performance. If you're using iSCSI or Fibre Channel for shared storage, ensure that multipath I/O (MPIO) is properly configured to provide redundancy and optimal performance.
 
 - At least two network interfaces per node are recommended, one for cluster communication and one for client access. Ensure that the network settings allow for proper communication between nodes, including correct IP addresses and subnet configurations. Make sure firewalls are configured to permit necessary cluster traffic.
 

WindowsServerDocs/storage/refs/refs-overview.md

@@ -142,7 +142,7 @@ Contact application and storage array vendors for support details on Fiber Chann
 | Page file support | ✅ | ✅ |
 
 1. Available on Windows Server, version 1709 and later, Windows Server 2019 (1809) LTSC or later.
-2. CSV won't use Direct I/O with Storage Spaces, Storage Spaces Direct (S2D), or SAN.
+2. ReFS-formatted CSVs don't use Direct I/O with Storage Spaces, Storage Spaces Direct (S2D), or SAN. For SAN storage, format CSVs with NTFS to use Direct I/O.
 3. Available on Windows Server 2025 only.
 4. Storage Spaces only.