Skip to main content

Druva Cloud Platform Backups: Understanding Irregular Daily Snapshots and Missing Days

Updated over a week ago

Problem Statement:

Admins observe that backups for various Druva Cloud Platform workloads aren't consistently occurring every calendar day, leading to "missing" daily snapshots and an inability to restore data from a specific date. Backup times appear irregular.

For example, in the image below, we can see a clear irregularity in backups as the entry for May 25, 2025, is missing:

Overview:

Druva's "Once a day" backup frequency for all supported workloads is designed to initiate a new backup 24 hours after the completion of the previous backup, not at a fixed time each calendar day. This means if a backup starts late in the evening and finishes early the next morning, the subsequent Daily backup will be scheduled for 24 hours after that completion time, effectively "skipping" a calendar day's snapshot in the Daily sequence. This behavior can also be exacerbated by factors such as:

  1. API Throttling: For cloud-based workloads (e.g., M365, Salesforce), the respective cloud service provider can throttle API requests, especially during peak times or for large data sets. This can significantly extend backup completion times.

  2. Large Changed Data Volumes: If there's a substantial amount of new or changed data to back up within a workload, the backup process will naturally take longer to complete.

Impact:

  • A specific day's backup snapshot might be unavailable for a given workload, preventing point-in-time restores for that particular day.

  • Perceived gaps in the Daily backup history.

Diagnosis:

To understand the backup pattern for any workload, check the activity logs for the specific backup job or data source. Note the completion time of each backup. The next backup should initiate approximately 24 hours after that recorded completion time.

  • Scenario Example:

    • If a workload backup (e.g., a SharePoint site, an Exchange Online mailbox, a Salesforce instance) is set for "Once a Day". Please refer to the below image:

  • The backup on May 20th, 2025, starts at 10:00 PM and successfully completes on May 20th, 2025, at 11:53 PM.

  • Since the new backup is scheduled 24 hours after completion, the next backup will then start on May 21st 2025, at 11:53 PM, however the backup ends on 22nd May, at 12:03 AM.

  • In this scenario, there will be no backup snapshot available for May 21st, 2025. Any data changes made on May 21st would be captured in the May 22nd backup.

Please refer to the below image:

Resolution/Workaround:

  1. Understand Druva's 24-Hour Backup Logic: Be aware that "Daily" in Druva's scheduling refers to a 24-hour interval from the previous backup's completion, not a fixed Daily window (e.g., midnight to midnight).

  2. Adjust Backup Frequency (Consider "Twice a Day"): If Daily restore points are critical, consider changing the backup interval to "Twice a day" where supported by the workload. This increases the frequency and might help cover potential gaps, though it doesn't eliminate the issue if large data changes or throttling continue to prolong backup windows.

  3. Check Later Snapshots for Data: If a specific day's snapshot is missing, the data you're looking for from that period will likely be included in the next available snapshot (e.g., if July 2nd was skipped, data from July 2nd would be in the July 3rd snapshot).


    πŸ“ Note

    This behavior is by design in Druva's current architecture for "Once a day" backups across all workloads.

Did this answer your question?