Hello igh08

List below is how we recommend you want to deploy your Powerstore 3000T and SQL Server2022 and Hyper-V

1. Dell PowerStore 3000T Configuration

 

The PowerStore 3000T provides dual-node (active/active) processing with end-to-end NVMe.

 

Connectivity Protocol:

Fibre Channel (FC) / NVMe-oF (NVMe/FC or NVMe/TCP): Recommended for lowest latency and highest throughput.

iSCSI: Ensure dedicated 10/25 GbE network interfaces with Jumbo Frames (MTU 9000) and separate VLANs for each path.

Multipathing (MPIO):

Configure Microsoft MPIO on Hyper-V hosts.

Use the Round Robin with a subset path policy or follow the Dell PowerStore Host Configuration Guide for Windows to optimize I/O pathing across both appliance nodes.

Volume Groups:

If a SQL Server database spans multiple volumes (e.g., separate volumes for Data, Logs, and TempDB), place them into a single PowerStore Volume Group with Write-Order Consistency enabled. This ensures crash-consistent array-level snapshots.

Volume Performance Policy:

Set the performance policy to High for production SQL Server data and transaction log volumes to prioritize critical I/O during periods of contention.

Data Reduction Considerations:

PowerStore features inline deduplication and compression that is always active.

SQL Server Native Compression / TDE:

If SQL Server Transparent Data Encryption (TDE) is used, enable SQL 2022’s optimized TDE or be aware that host-encrypted data reduces the efficiency of storage-array deduplication and compression.

 

2. Hyper-V Configuration Best Practices

Cluster Shared Volumes (CSV):

Deploy CSVs for simplified management and live migration support.

Format CSV volumes with ReFS or NTFS using a 64 KB allocation unit size.

Virtual Machine Architecture:

Use Generation 2 virtual machines for modern UEFI support, faster boot, and improved virtual device performance.

Static Memory Allocation: Do not use Dynamic Memory for production SQL Server instances. Allocate static RAM to prevent performance degradation and memory thrashing.

vCPU and NUMA: Configure virtual sockets and virtual cores to match the physical host's NUMA architecture. Ensure the VM does not span across physical NUMA nodes unless the workload strictly requires more memory/CPUs than a single NUMA node can provide.

Virtual Disk Layout (VHDX):

Use VHDX formatted disks (Fixed Size is recommended for mission-critical write-heavy workloads to avoid runtime expansion overhead).

Distribute virtual disks across multiple Synthetic SCSI Controllers inside the VM (e.g., SCSI 0 for OS, SCSI 1 for Data, SCSI 2 for Logs, SCSI 3 for TempDB) to maximize parallel queue depths.

 

3. SQL Server 2022 Best Practices

 

Disk Partition Formatting:

Format all database volumes inside the guest OS (Data, Log, TempDB) with an allocation unit (cluster) size of 64 KB.

Drive Separation:

Operating System: Dedicated VHDX.

User Data Files (.mdf / .ndf): Separate VHDX mapped to a dedicated PowerStore volume.

Transaction Logs (.ldf): Separate VHDX mapped to a dedicated PowerStore volume (optimized for sequential writes).

TempDB: Dedicated VHDX.

TempDB Sizing:

Allocate multiple TempDB data files (typically 1 file per vCPU up to 8; add in increments of 4 if allocation contention is observed).

Ensure all TempDB data files are pre-sized identically.

SQL Server Instance Settings:

Max Server Memory: Cap SQL Server maximum memory to leave sufficient headroom (typically 4–8 GB or more, depending on host tasks) for the OS and virtualization overhead.

Instant File Initialization (IFI): Grant the SQL Server service account the `Perform Volume Maintenance Tasks` privilege to speed up data file allocations and restores.