Atlas Data Federation single-collection proxy optimization
Parent: MongoDB Atlas Data Federation · Published reference · snapshot 2026-10-29
↓ Facts as markdownall context files
Source-anchored research for Atlas Data Federation single-collection proxy optimization. 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.
Structure and components
- Partial pushdown is a separately described optimization: Data Federation tries to execute eligible stages such as $match in Atlas and returns their results to the federated layer for subsequent stages. It is not a promise that every pipeline executes wholly in Atlas. (low confidence) [source]
How it works
- The single-collection proxy optimization is unavailable when the queried virtual collection maps to multiple Atlas collections; one Atlas cluster alone is not the documented eligibility condition. (low confidence) [source]
- The proxy optimization is unavailable when the queried virtual collection maps to multiple Atlas collections, even if those collections belong to the same Atlas store. (low confidence) [source]
Comparisons and alternatives
- Object-store partition pruning selects files using query parameters and configured partition attributes. This file-selection mechanism differs from forwarding a query to a single Atlas collection. (medium confidence) [source]
- In Stack Overflow comments, a user reported a 30+ second response through Atlas Data Federation versus at most one second when querying the source directly. Another participant listed query shape, network or topology differences, and a bug as possible contributors; the exchange does not establish which cause applied. (low confidence) [source]
Facts and statements
- A Data Federation collection mapped to only one Atlas collection acts as a proxy: it forwards the query to Atlas without scanning data into its virtual collection. (low confidence) [source]
- MongoDB describes the proxy path as improving performance and reducing cost, but its explanation supplies no speedup factor, cost percentage, or zero-charge guarantee. (low confidence) [source]
- The proxy path avoids scanning data into the virtual collection for query processing; MongoDB describes improved performance and reduced cost without quantifying either benefit. (low confidence) [source]
- MongoDB's configuration example places foo and barbaz in one federated instance: foo has one Atlas collection mapping and receives proxy optimization; barbaz has two mappings and does not. (low confidence) [source]
- The documented mechanisms differ: single-collection proxying avoids virtual-collection scanning, whereas object-store partition pruning selects files using query parameters. The latter does not establish proxy eligibility. (medium confidence) [source]
Children
- No children recorded.