Backup and PITR cost modelling
Parent: Atlas Cost Planning: Region Pricing, Estimation, Scheduled Scaling and Competitor TCO · Published reference · snapshot 2026-10-24 · skill mongodb-cohort-c8ad8a77bff0
↓ Facts as markdownall context files
Source-anchored research for Backup and PITR cost modelling. Run status: COMPLETED.
These notes link each claim to its source. Some citations identify archived research reports that are no longer hosted on this site. A published reference means the content is available; it does not certify independent review or accuracy.Read the editorial policy and follow the sources before relying on a claim.
Definitions
- Dedicated-cluster backup storage is a separate cost dimension from instance price; ordinary snapshot rates vary by provider and region. (high confidence) (verified as of 2026-10-24) [source]
- nilDB's admin guide says PITR is not required for its deployment and directs Atlas users to disable Continuous Cloud Backups while retaining hourly snapshots for one day and daily snapshots for seven days. Heureka reports its own restore test from the latest backup and says disaster-recovery plans should be tested regularly. These are deployment-specific policy and operator reports, not a universal backup requirement. (medium confidence) (verified as of 2026-10-24) [source]
Measurements and reference values
- Heureka's article dated 24 March 2025 gives an example of about $30/month for Atlas backups every six hours with two-day retention. In a restore test, restoring from the latest backup took 8 minutes 5 seconds on an M10 cluster with 1.7 GB RAM, 40 GB storage and 2,400 IOPS. This is a workload-specific observation, not a current Atlas price quote or a general RTO guarantee. (low confidence) (verified as of 2025-03-24) [source]
- The documented AWS us-east-1 example prices 115 GB of combined snapshot and oplog storage at $106.25/month: the first 5 GB are free, then 95 GB at $1 and 15 GB at $0.75. This is a regional tiered example, not a universal backup rate. (high confidence) (verified as of 2026-10-24) [source]
Problems, failure modes and limitations
- Treat compliance minimums as constraints on backup cost optimisation. Cluster policies cannot undercut a project Backup Compliance Policy; reducing a protected PITR window requires MongoDB Support and verification by the designated security or legal representative. (high confidence) (verified as of 2026-10-14) [source]
Comparisons and alternatives
- Reconcile snapshot storage using GB-days: the documented daily rate is monthly GB rate multiplied by 12/365. Repeated daily invoice rows represent retained storage over time, rather than inherently duplicated charges. (high confidence) (verified as of 2026-10-24) [source]
- At AWS us-east-1 example rates, 115 GB costs $16.10/month as ordinary snapshots versus $106.25/month as combined continuous-backup storage. This equal-byte comparison excludes additional oplog growth. (high confidence) (verified as of 2026-10-24) [source]
- Failover, configuration changes and cloud-provider maintenance can cause extra full snapshots. Backup storage may reflect provisioned volume capacity rather than live data size. (high confidence) (verified as of 2026-10-24) [source]
- Atlas Infinite includes a continuous 24-hour backup window. Additional retention, enabled by default, can incur extra backup charges; backup storage is metered hourly in uncompressed GB-hours. The backupEnabled API field controls additional retention rather than the included window. (high confidence) (verified as of 2026-10-20) [source]
Facts and statements
- For Atlas Core, PITR storage billing uses snapshots plus the internal oplog under the Continuous Cloud Backup schedule. A longer PITR window increases oplog storage; model its cost separately from snapshot retention duration. (high confidence) (verified as of 2026-10-24) [source]
- Do not estimate retained storage as live database size multiplied by snapshot count. Incremental snapshots share unchanged data, while retained history and write churn affect storage; deleting a snapshot need not release every byte it references. (high confidence) (verified as of 2026-10-24) [source]
- Retention savings require checking existing snapshots: Terraform update_snapshots explicitly applies updated retention to previously taken snapshots; changing future policy alone can leave old retention costs. (high confidence) (verified as of 2026-10-24) [source]
- Budget backup separately from cluster configuration and data transfer. AWS migration guidance lists backup as a distinct Atlas cost category; MongoDB documents separate backup storage and restore-related charges. (high confidence) [source]
- Savings forecasts must distinguish future policy changes from existing snapshots. Applying shorter retention to existing eligible snapshots requires the explicit Update retention time of existing snapshots option; otherwise old retained snapshots can delay savings. (high confidence) (verified as of 2026-10-14) [source]
- PointFive recommends backup tiers by environment and criticality and reducing snapshot frequency or retention for less-critical clusters only where recovery objectives allow. OhChimp demonstrates inspecting Atlas invoice backup line items, the cluster backup schedule, `pitEnabled`, and snapshot storage sizes to analyze backup charges. (medium confidence) [source]
- MongoDB reported a bug that inflated backup charges for the April 24, 2026 billing date. Its April 27 resolution said corrected usage should appear by April 28. Treat unexpected jumps as a reconciliation investigation before attributing them to workload growth. OutageDeck mirrors MongoDB statements and is not an independent incident origin. (low confidence) (verified as of 2026-04-27) [source]
Children
- No children recorded.