Sensitive Virtual Machines become impacted by snapshot with memory/snapshot consolidation stun time
search cancel

Sensitive Virtual Machines become impacted by snapshot with memory/snapshot consolidation stun time

book

Article ID: 444813

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

  • Applications running inside a Virtual Machine may crash or fail unexpectedly.

  • The virtual machine will briefly not respond to network pings.

  • Highly sensitive workloads, such as database or banking applications, crash or experience severe performance degradation.

Environment

  • VMware vSphere ESXi 7.x

  • VMware vSphere ESXi 8.x

  • VMware vSphere ESXi 9.x

Cause

When a virtual machine snapshot is created with the virtual machine memory state included, the ESXi host must stun the virtual machine to write its active memory contents to disk.

Virtual machines undergo a brief stun (pause) during snapshot creation and snapshot consolidation processes. While the majority of Guest Operating Systems and applications are not affected by this stun, highly sensitive applications like databases or financial systems may crash or experience persistent performance impacts due to the stun time required to quiesce the disk or commit the snapshot delta disks.

Resolution

  1. Deselect the "Snapshot the virtual machine's memory" option when taking manual snapshots of highly sensitive virtual machines.
  2. Configure third-party backup software to exclude the virtual machine memory state when requesting snapshots.
  3. Schedule virtual machine backups or manual snapshot operations exclusively during off-peak hours or designated maintenance windows.
  4. If the application is highly sensitive to compute or storage stuns, configure backup software to utilize storage array-level snapshots instead of virtual machine-level snapshots.
  5. Consolidate snapshots regularly to prevent large delta disks from building up, which significantly increases the stun duration required for consolidation.
  6. Contact the Guest OS or Application vendor for tuning recommendations to help the application tolerate brief infrastructure stuns without crashing.

Additional Information