Visitar URL original
docs: Add MongoDB data-source and offline-store reference documentation by jvincent-mongodb · Pull Request #6351 · feast-dev/feast · GitHub
Skip to content

docs: Add MongoDB data-source and offline-store reference documentation #6351

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ntkathole merged 11 commits into from
May 1, 2026
Merged

docs: Add MongoDB data-source and offline-store reference documentation #6351

Changes from 1 commit
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
b322422
docs: Add MongoDB docs for data-source and offline-store integration
jvincent-mongodb Apr 29, 2026
506d235
docs: various copy revisions
jvincent-mongodb Apr 29, 2026
e1ab8d8
docs: Update MongoDB data-source documentation with Vector Search sec…
jvincent-mongodb Apr 30, 2026
03aaa78
docs: fix example import path for MongoDB data-source
jvincent-mongodb Apr 30, 2026
71b8853
docs: fix broken url
jvincent-mongodb Apr 30, 2026
3eab42a
docs: fix code example
jvincent-mongodb Apr 30, 2026
8d6125a
docs: minor edits to MongoDB data-source and offline-store docs
jvincent-mongodb Apr 30, 2026
6f0af71
Merge branch 'master' into master
jvincent-mongodb May 1, 2026
c6f1223
docs: minor docs edits
jvincent-mongodb May 1, 2026
a23e7e2
Merge branch 'master' into master
jvincent-mongodb May 1, 2026
49ab239
docs: minor docs edits
jvincent-mongodb May 1, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
docs: various copy revisions 
Signed-off-by: jvincent-mongodb <jeffrey.vincent@mongodb.com>
  • Loading branch information
@jvincent-mongodb
jvincent-mongodb committed Apr 30, 2026
commit 506d235fa44b4cc5a151ee89b8b5ebe20bd91701
8 changes: 5 additions & 3 deletions docs/reference/offline-stores/mongodb.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,8 +3,8 @@
## Description

The MongoDB offline store provides support for reading [MongoDBSource](../data-sources/mongodb.md).
* Entity dataframes can be provided as a Pandas dataframe. The offline store converts entity identifiers into serialized entity keys for efficient lookup against the `feature_history` collection.
* Uses a single shared `feature_history` collection with a compound index for all FeatureViews, distinguished by a `feature_view` discriminator field.
* Entity dataframes can be provided as a Pandas dataframe. The offline store converts entity identifiers into serialized entity keys for efficient lookup against the `feature_history` collection.

@caseyclements caseyclements May 1, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Skip the description.

## Getting started

Expand All @@ -26,6 +26,8 @@ online_store:
type: mongodb
connection_string: "mongodb+srv://user:pass@cluster.mongodb.net"
Comment thread
jvincent-mongodb marked this conversation as resolved.
Outdated
database_name: feast_online_store
collection_suffix: latest
client_kwargs: {}
```
{% endcode %}

Expand Down Expand Up @@ -66,8 +68,8 @@ This index enables efficient range scans over entities and feature views, while
## Key Optimizations

* **K-collapse**: Multiple FeatureViews that share the same join keys are queried in a single aggregation using `feature_view: {$in: [...]}`, reducing round trips.

@ntkathole ntkathole May 1, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The actual get_historical_features implementation loops over each proj_name in fv_to_features and issues separate MongoDB aggregation pipelines per feature view.

K-collapse can be enhancements?

@caseyclements caseyclements May 1, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You're right. K-collapse was removed after the complication around projection keys was discovered. Two projections, same feature view. Thank you for pointing this out. I may have to change docstrings too.

* **Scoring vs. training paths**: When `entity_df` has unique entity IDs (scoring), server-side `$group $first` is used for efficient retrieval. When entity IDs repeat (training), `pd.merge_asof` provides correct point-in-time joins.
* **Two-level chunking**: Outer `CHUNK_SIZE` (50,000 rows) limits the entity DataFrame slice; inner `MONGO_BATCH_SIZE` (10,000 entity IDs) limits `$in` array size per aggregation call.
* **Scoring vs. training paths**: When each entity appears only once in `entity_df` (scoring/inference — one feature lookup per entity), server-side `$group $first` efficiently returns the single latest value per entity. When the same entity appears at multiple timestamps (training — building a dataset with many historical snapshots per entity), the store retrieves all candidate rows and uses `pd.merge_asof` to select the correct point-in-time value for each request timestamp.
* **Two-level chunking**: `CHUNK_SIZE` (50,000 rows) controls the size of intermediate DataFrames in memory; `MONGO_BATCH_SIZE` (10,000 entity IDs) limits the query size sent to MongoDB.

## Functionality Matrix

Expand Down