All VMs experience drops in pings when a snapshot is being taken or removed. This is causing issues with sensitive production SQL servers, job failures, and DB disconnects.

Products

• VMware vSAN

Issue/Introduction

During snapshots of VMs experiencing ping drops, disconnects, stun time.

This results in job failures for SQL VMs and database disconnects

 

Resolution

In general, snapshot usage should be limited and possibly avoided on VMs running production SQL Server workloads. For backup purposes, consider to use in-guest, agent-based backup. If snapshots should be taken, the following best practices should be considered:

• Offline snapshot (a VM is powered off when a snapshot is taken) can be used without special considerations.
• If an online snapshot (VM is powered on and Guest OS is running) needs to be taken:

– Consider not using “Snapshot the virtual machine’s memory” option as this may stun a VM. Rely on SQL Server mechanisms to prevent data loss by losing in-memory data.
– Use “Quiesce guest file system” option to ensure that a disk consistent snapshot will be taken.

Special notes:
» Consider not taking an online snapshot if VMware Tools are not installed or not functional as this may lead to the inconsistent disk state.

Please review the below PDF, particularly page 59 for snapshots of SQL VMs.
ARCHITECTING MICROSOFT SQL SERVER ON VMWARE VSPHERE