Storage layout and best practices for Amazon RDS Custom for SQL Server

This article explains the storage architecture of Amazon RDS Custom for SQL Server and provides guidance on safely managing internal storage volumes to prevent data loss.

• The W: drive is a block-level replication layer beneath the D: data volume; formatting it causes complete data loss
• Multi-AZ deployments have C: (OS), D: (data), and W: (replication); secondary instances have only C: and W:
• W: appears as RAW in Disk Management, which is normal and expected; never format, partition, or modify it
• Use the provided PowerShell script to identify drives by EBS volume ID before any disk operation
• Only attach new user-managed volumes with proper tagging and verify size and EBS ID before formatting
• Exclude W: and D: from maintenance scripts, defragmentation tools, and disk cleanup utilities
• Real customer incidents include mistaking W: for new volumes, running defragmentation on replication partitions, and automating operations without drive verification
• If W: is accidentally formatted, pause RDS Custom automation and restore from the most recent backup

Understanding the C:, D:, and W: drive relationships and treating W: as off-limits protects production data when using OS-level access in RDS Custom for SQL Server.