Relational-to-MongoDB migration cost (Relational Migrator)
Parent: Migration and Cutover Cost (Live Migration, mongosync, Dual-Running, Egress) · Published reference · snapshot 2026-10-31
↓ Facts as markdownall context files
Source-anchored research for Relational-to-MongoDB migration cost (Relational Migrator). 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.
How it works
- The cached MongoDB overview, fetched October 14, 2026, says Migrator supports only snapshot jobs. Its scheduled-downtime procedure permits source reads but stops source writes during migration. Repeated scheduled snapshots should not be budgeted as continuous change capture without another mechanism. (low confidence) (verified as of 2026-10-14) [source]
Measurements and reference values
- Doubling the Migrator host from 16 vCPUs/32 GB RAM to 32 vCPUs/64 GB RAM did not halve runtime: MongoDB reports 0% improvement for 10 GB one-to-one and 4.82% for 100 GB advanced mapping. Cost estimates should measure bottlenecks before assuming larger hosts pay for themselves. (low confidence) [source]
- MongoDB's v1.15 benchmark used an isolated Oracle RDS source and a 32-vCPU, 64-GB Migrator instance. For 100 GB, one-to-one mapping took 2h31m on M60, embedded mapping 16h25m on M80, and advanced mapping 15h47m on M140; these are configuration-specific observations, not fixed migration prices. (low confidence) [source]
- MongoDB's benchmark guidance gives approximately 21 GB/hour for one-to-one mapping and 7 GB/hour for up-to-two-level embedded mappings. The same page's 100-GB one-to-one table row implies about 40 GB/hour, so the planning estimate and measured row should not be presented as interchangeable guarantees. (low confidence) [source]
Comparisons and alternatives
- In MongoDB's Relational Migrator v1.15 benchmark, 100 GB took 2 hours 31 minutes with one-to-one mapping on M60, versus 16 hours 25 minutes with embedded mappings on M80. The Migrator host had 32 vCPUs and 64 GB RAM. These are workload-specific observations, not migration-time guarantees. (low confidence) [source]
- Adiom's October 28, 2025 vendor comparison reports a 30 GB PostgreSQL TPC-H transfer in 40 minutes or less with Dsync, while its Relational Migrator runs took 6–8 hours and ended in a connection error. It does not identify the Migrator version, and its dataset, infrastructure and mapping differ from MongoDB's Oracle benchmark; its speedup claim is not independently established. (low confidence) [source]
- Schema transformation belongs in migration cost estimates: MongoDB reports nonlinear scaling for complex embedded mappings, while Adiom's SQL-batch experiment reshapes data with source-side SQL joins and aggregation. These support testing the actual mapping rather than estimating runtime from source bytes alone. (medium confidence) [source]
- Adiom's October 28, 2025 competitor-authored test reports Dsync initial sync in 20–40 minutes for a 30-GB TPC-H dataset, versus 6–8 hours followed by a connection failure for Relational Migrator. The Migrator version is unspecified, and the post describes both sharded and replica-set targets, limiting reproducibility and generalization of its claimed 10–20x advantage. (low confidence) [source]
Facts and statements
- MongoDB describes Relational Migrator as a free tool; this establishes tool availability without a license charge, not a zero-cost migration project. (low confidence) (verified as of 2026-10-14) [source]
- Migration planning must budget work beyond copying rows: AWS lists working-set sizing, bandwidth estimation, Atlas tier selection, security configuration and application cutover; MongoDB's benchmark separately requires source, Migrator-host and target infrastructure. This is a cost-work breakdown, not a measured dollar estimate. (medium confidence) [source]
- In a May 10, 2024 profiling example, a MongoDB Relational Migrator engineer reports 120 seconds spent writing to MongoDB within a 130-second migration, with 3.5 seconds in migration logic. This supports investigating destination I/O in that run, not assuming the same proportions for all migrations. (low confidence) [source]
- Relational Migrator can serve as a modeling step independently of its transfer engine: MongoDB documents visual schema mapping, and Adiom demonstrates exporting a .relmig project to generate a SQL-batch connector configuration. Choosing another transfer engine can therefore retain the mapping work, although these sources do not quantify labor saved. (medium confidence) [source]
- AWS's implementation pattern estimates approximately one week for database sizing and typically one day for bandwidth estimation. These are planning suggestions for named tasks, not measured end-to-end migration duration or a universal staffing estimate. (low confidence) [source]
- MongoDB describes Relational Migrator as a free schema-mapping and snapshot-migration tool. Free licensing should not be treated as zero project cost: AWS's implementation pattern separately requires sizing, networking, security configuration and application cutover work. (medium confidence) (verified as of 2026-10-14) [source]
- Relational Migrator 1.15, released October 17, 2025, moved continuous sync (CDC), LLM-assisted query conversion and entity code generation to Application Modernization Platform engagements. The overview describes the standalone tool as snapshot-only; a current online-migration estimate must account for a separate synchronization approach or engagement. (high confidence) (verified as of 2025-10-17) [source]
- A migration budget should include representative rehearsals and temporary infrastructure sizing. MongoDB recommends monitoring source, Migrator and target bottlenecks; AWS independently recommends benchmarking table migrations and resizing or deleting idle migration instances. AWS's guidance concerns DMS and supplies no Relational Migrator-specific savings percentage. (medium confidence) [source]
- AWS's relational-to-Atlas pattern assigns working-set assessment, network estimation, cluster selection and cutover planning to DBAs and application owners. These are explicit implementation tasks beyond copying rows; the pattern does not provide a measured dollar cost for a completed migration. (low confidence) [source]
Children
- No children recorded.