Growth Signals
Parent: MongoDB Capacity Planning · Published reference · snapshot 2026-10-23 · skill mongodb-cohort-1d798a0d85d1
↓ Facts as markdownall context files
Source-anchored research for Growth Signals. 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
- For Atlas Search, monitor search-index size separately from source collection size because mappings and features can change their size ratio. MongoDB recommends establishing a normal index-size and growth-rate baseline, watching unexplained jumps or sustained growth above that rate and declining free disk, and reserving room for the largest index plus expected growth and rebuilds. (high confidence) (verified as of 2025-08-12) [source]
- After documents are removed or shrink, dataSize can decrease while storageSize does not. Compare freeStorageSize and indexFreeStorageSize, when collected, to identify allocated but unused space; PMM's fragmentation panel likewise describes unused allocated storage after documents are deleted or moved. (high confidence) [source]
- An index-size trend that rises faster than data size is a useful diagnostic signal. Percona notes that it can indicate indexes accumulating without matching document growth, for example when TTL indexes are not cleaning up data as expected or old indexes remain; MongoDB exposes allocated index size separately as dbStats.indexSize. (high confidence) [source]
- Track average document size over time as well as document count: MongoDB defines avgObjSize as dataSize divided by object count, and Percona notes that a rising average can indicate documents themselves are growing, such as when arrays are appended. (medium confidence) [source]
Measurements and reference values
- For DigitalOcean Managed MongoDB, the documentation recommends keeping disk usage below 90% and notes that vacuuming can cause fluctuations, especially for write-heavy use cases; treat this as provider-specific guidance. (low confidence) (verified as of 2026-07-31) [source]
Facts and statements
- Compare total data-plus-index size over time against a normal growth baseline and expected storage runway. Percona's MongoDB ReplSet Summary dashboard recommends using this trend to plan storage additions and investigate unexpected spikes; MongoDB's manual says an unexpected change in dataSize, indexSize, storageSize, or document count while traffic remains ordinary can indicate data growth or index changes. (high confidence) [source]
- Treat logical data size and physical disk consumption as different signals: Atlas defines dataSize as uncompressed logical data and db index size as disk space allocated to indexes, including free index space, while Disk Space Used and Disk Space Free describe physical consumption and headroom. WiredTiger compression can make logical data and on-disk size differ, so track both database-size metrics and actual free disk. (high confidence) [source]
- Watch memory and cache pressure alongside storage growth. Percona's dashboard compares WiredTiger cache use with its maximum and says sustained near-full use increases eviction; sustained page faults can mean the working dataset exceeds host memory and must be read from disk. MongoDB's manual also treats rising page faults or growing data changed but not yet written as signs of memory pressure. (high confidence) [source]
- DigitalOcean Managed MongoDB's Insights graphs cover 1 hour to 14 days, may have gaps around outages, maintenance, failovers, or migrations, and disk usage can fluctuate during vacuuming. This limits what one chart alone can establish about longer-term growth. (low confidence) (verified as of 2026-07-31) [source]
- Ops Manager records the source database's data size, storage size, and file size at completed snapshots, and tracks snapshot-store live, dead, and total bytes over time. The graph updates periodically, so recent changes may lag; treat backup-store growth as a separate capacity signal from source database growth. (low confidence) (verified as of 2026-10-14) [source]
- MongoDB's October 2026 engineering account describes an accelerating large Atlas cluster where an internal estimate for storage expansion took several days. This is one case, but it illustrates why planners should compare growth runway with the time required to expand storage. (low confidence) [source]
- Track logical data, collection allocation, and index allocation as separate size series. MongoDB defines dbStats.dataSize as uncompressed data, storageSize as collection storage allocation including free space but excluding indexes, and indexSize as allocated index space; under WiredTiger compression, dataSize may exceed storageSize. Percona also charts logical data and index size separately. (high confidence) [source]
- Monitor filesystem headroom separately from database size. MongoDB dbStats reports fsUsedSize and fsTotalSize for the filesystem holding MongoDB data; PMM charts filesystem space use and warns that a full disk can stop MongoDB from accepting writes, while DigitalOcean's managed MongoDB metrics show minimum, maximum, and average disk use and note vacuum-related fluctuations. (high confidence) [source]
Children
- No children recorded.