Atlas Search Nodes cost floor and co-located versus dedicated search
Parent: Cluster Tier Limits and Feature Gates as Cost Drivers (Connections, IOPS Caps, Minimum Tiers) · Published reference · snapshot 2026-10-31
↓ Facts as markdownall context files
Source-anchored research for Atlas Search Nodes cost floor and co-located versus dedicated search. 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
- Dedicated Search Nodes do not make Atlas Search synchronously consistent with database writes. MongoDB's Erik Hatcher describes eventual-consistency lag as dependent on workload, architecture and resource contention; his design-review example used a separate cache when lag did not meet the application's requirement. (low confidence) [source]
- MongoDB documents a minimum of two dedicated Search Nodes, billed hourly per node. Atlas bills partial hours as full hours, starts charging when the configuration is created, and bills both instance sizes for the first hour after a search-tier change. (low confidence) (verified as of 2026-10-31) [source]
Measurements and reference values
- Dedicated Search Nodes deploy per shard: MongoDB's example assigns six Search Nodes when two are configured for each of three shards. For a single-region configuration with uniform hourly node price p and S shards, the documented minimum implies search compute of 2 × S × p per billed hour, before database and network charges; this is a derived formula, not a vendor dollar quote. (high confidence) (verified as of 2026-10-31) [source]
Problems, failure modes and limitations
- Co-located search has availability exposure during cluster changes: MongoDB documents cases where mongod is available before mongot and search queries fail. An October 12, 2026 user test reported 55 final HTTP 503 responses during a scale-down despite retries across three members; that observation is one deployment, not a fleet-wide failure rate. (medium confidence) [source]
Comparisons and alternatives
- Atlas bills a full hour even for partial-hour Search Node use. On an instance-size change, it charges for both old and new instances for the first hour, then only the new instance. Replacement of an unhealthy node is charged as one node rather than two. (low confidence) (verified as of 2026-10-31) [source]
- Dedicated Search Node costs include node count, search tier and traffic between Search Nodes and database nodes. Network-transfer charges appear at cluster level rather than as a separate Search Node network line, so a TCO model must include them. (high confidence) (verified as of 2026-10-31) [source]
- MongoDB recommends dedicated Search Nodes for production; this is deployment guidance rather than proof that co-located search is unavailable. Its co-location guidance warns that contention can impair indexing and query latency, while Tidewave independently describes the same resource-sharing trade-off. (medium confidence) (verified as of 2026-10-31) [source]
Facts and statements
- MongoDB's deployment documentation says co-located Atlas Search adds no separate enablement fee, but mongod and mongot share resources; large indexes can increase cluster resource utilization. 'No additional search fee' therefore does not establish zero incremental infrastructure cost. (medium confidence) (verified as of 2026-10-31) [source]
- The cached MongoDB deployment and billing documentation requires a minimum of two dedicated Search Nodes, on an M10-or-higher database cluster, with hourly charges per Search Node. This is an additional search-capacity floor alongside the database deployment. (high confidence) (verified as of 2026-10-31) [source]
- Dedicated Search Nodes separate mongot from database-serving nodes and allow search capacity to scale independently. This removes direct mongod/mongot resource contention on the same host, providing a capacity-management benefit that must be weighed against additional node costs. (high confidence) [source]
- Baris Hantas's May 20, 2025 implementation account reports retaining three standard database nodes and adding two Search Nodes, with stable performance under concurrency. Treat this as an attributed workload-specific account: the post does not supply a controlled general price comparison between co-located and dedicated configurations. (low confidence) [source]
- Atlas integration automates search-index synchronization and retains the MongoDB aggregation API in both deployment patterns. Tidewave and Tapptitude describe reduced synchronization and integration work; those operational benefits do not establish a measured infrastructure savings percentage. (medium confidence) [source]
Children
- No children recorded.