Event details

Take a closer look at how storage is evolving in Windows Server, from NVMe improvements to high‑level platform investments. Dive into the details of native NVMe support, NVMe‑OF direction, ReFS enhancements, and Storage Spaces and S2D improvements. We’ll walk through what’s available today in Windows Server 2025, highlight recent performance and resiliency enhancements, and share what’s coming next in future versions of Windows Server.

 

This session is part of Windows Server Summit 2026.

Pearl-Angeles
Updated May 15, 2026

6 Comments

Comments have been turned off for this event
  • Thanks for tuning in! We hope you enjoyed this session and Windows Server Summit 2026. We’ll continue to answer questions here in the chat through the end of the week. If you missed the live broadcast, don’t worry – you can watch it on demand.

    In fact, you can catch any of this week’s great Windows Server Summit 2026 sessions on demand by visiting https://aka.ms/WindowsServerSummit.

    Share your feedback on Windows Server Summit. Click the image below to take our very short survey - thank you! 👇

  • jpeg's avatar
    jpeg
    Copper Contributor

    Will REFS be updated to support SAN based CSVs?  Currently when using ReFS on SAN-based CSVs, the volume operates in File System Redirected Mode, meaning all I/O passes through the coordinator node over SMB rather than direct block-level access. This can cause severe performance degradation compared to NTFS, which supports Direct I/O in SAN environments.

    • Thank you for sharing this feedback. We are aware of this limitation and are investigating solutions to this for Windows Server vNext.

  • AKFurman's avatar
    AKFurman
    Copper Contributor

    Hello, Just curios, would there be any effect on drives lifespan using native NVMe?

    • Yash_Shekar's avatar
      Yash_Shekar
      Icon for Microsoft rankMicrosoft

      Native NVMe doesn’t change SSD wear characteristics directly—it just removes inefficiencies in the OS storage stack. Drive lifespan is primarily driven by total writes and write amplification. The only way lifespan would change is indirectly, for example if higher performance leads to higher sustained write throughput in real workloads.