Vertex AI Feature Store is a managed, cloud-native feature store servicethat's integral to Vertex AI. It streamlines your ML feature managementand online serving processes by letting you manage your feature data in aBigQuery table or view. You can then serve features online directly from theBigQuery data source.

Vertex AI Feature Store provisions resources that let you set uponline serving by specifying your feature data sources. It then acts as a metadatalayer interfacing with the BigQuery data sources and servesthe latest feature values directly from BigQuery for onlinepredictions at low latencies.

In Vertex AI Feature Store, the BigQuery tables orviews containing the feature data collectively form theoffline store. You canmaintain feature values, including historical feature data, in the offline store.Because all the feature data is maintained in BigQuery,Vertex AI Feature Store doesn't need to provision a separateoffline store within Vertex AI. Moreover, if you want to use thedata in the offline store to train ML models, you can use the APIs andcapabilities in BigQuery to export or fetch the data.

The workflow to set up and start online serving using Vertex AI Feature Storecan be summarized as follows:

1. Prepare your data source in BigQuery.

2. Optional: Register your data sources by creating feature groups and features.

3. Set up online store and feature view resources to connect the feature data sourceswith online serving clusters.

4. Serve the latest feature values online from a feature view.

Vertex AI Feature Store data model and resources

This section explains the data models and resources associated with thefollowing aspects of Vertex AI Feature Store:

• Data source preparation in BigQuery

• Feature Registry setup

• Online serving setup

• Online serving

Data source preparation in BigQuery

During online serving, Vertex AI Feature Store uses feature datafrom BigQuery data sources. Before you set up Feature Registryor online serving resources, you must store your feature data in one or moreBigQuery tables or views.

Within a BigQuery table or view, each column represents a feature.Each row contains feature values corresponding to a unique ID. For moreinformation about how to prepare the feature data in BigQuery,seePrepare data source.

For example, in figure 1, the BigQuery table includes thefollowing columns:

• f1 andf2: Feature columns.

• entity_id: An ID column containing the unique IDs to identify each feature record.

• feature_timestamp: A timestamp column.

Because you prepare the data source in BigQuery and not in Vertex AI,you don't need to create any Vertex AI resources at this stage.

Feature Registry setup

After you've prepared your data sources in BigQuery, you canregister those data sources, including specific feature columns, inthe Feature Registry.

Registering your features is optional. You can serve features online even if youdon't add your BigQuery data sources to the Feature Registry. However,registering your features is advantageous in the following scenarios:

• Your data contains multiple instances of the same entity ID and you needto prepare your data in a time-series format with a timestamp column. When youregister your features, Vertex AI Feature Store looks up thetimestamp and serves only the latest feature values.

• You want to register specific feature columns from a data source.

• You want to aggregate specific columns from multiple data sources to definea feature view instance.

• You want to monitor the feature statistics and detect feature drift.

Caution: If you choose not to register your features, you must ensure thatevery row contains a unique value in theID column. For more information, seeData source preparation guidelines.

There are two types of Vertex AI Feature Store resources in theFeature Registry:

• Feature Registry resources for feature data

• Feature Registry resources for feature monitoring

Feature Registry resources for feature data

To register your feature data in the Feature Registry, you need to create thefollowing Vertex AI Feature Store resources:

• Feature group (FeatureGroup):AFeatureGroupresource is associated with a specific BigQuery source table orview. It represents a logical grouping of feature columns, which are represented byFeature resources.A feature group also contains one or multiple entity ID columns to identifythe feature records. If the feature data is in a time-series format, the featuregroup must also contain a timestamp column. For information about how to createa feature group, seeCreate a feature group.

• Feature (Feature):AFeatureresource represents a specific column containing feature values from the featuredata source associated with its parentFeatureGroup resource. For informationabout how to create features within a feature group, seeCreate a feature.

For example, figure 2 illustrates a feature group including feature columnsf1 andf2, sourced from a BigQuery table associated with thefeature group. The BigQuery data source contains four featurecolumns—two columns are aggregated to form the feature group. The featuregroup also contains an entity ID column and a feature timestamp column.

Feature Registry resources for feature monitoring

Feature monitoring resources let you monitor the feature data registered usingFeatureGroupandFeature resources. You can create the following resourcesrelated to feature monitoring:

• Feature monitor (FeatureMonitor):AFeatureMonitor resource is associated with aFeatureGroup resource and one or more featureswithin that feature group. It specifies the monitoring schedule. You can createmultiple feature monitor resources to set up different monitoring schedulesfor the same a set of features within a feature group. For example, if the featuresf1 andf2 are updated every hour, but the featuresf3 andf4 areupdated every day, you can create two feature monitor resources to efficientlymonitor these features:

  • Feature monitorfm1 that runs a monitoring job every hour on the featuresf1 andf2.

  • Feature monitorfm2 that runs a monitoring job every day on the featuresf3 andf4.

• Feature monitor job (FeatureMonitorJob):AFeatureMonitorJobresource contains the feature statistics and informationretrieved when a feature monitoring job is run. It can also contain information aboutanomalies, such as feature drift, detected in the feature data.

For more information about how to create feature monitoring resources, seeMonitor features for anomalies.

Online serving setup

To serve features for online predictions, you must define andconfigure at least one online serving cluster, and associate it with your featuredata source or Feature Registry resources. In Vertex AI Feature Store,the online serving cluster is called anonline store instance. An online storeinstance can contain multiplefeature view instances, where each feature viewis associated with a feature data source.

Online serving resources

To set up online serving, you must create the followingVertex AI Feature Store resources:

• Online store (FeatureOnlineStore):AFeatureOnlineStoreresource represents an online serving cluster instance and contains theonline serving configuration, such as the number of online serving nodes. Anonline store instance doesn't specify the source of the feature data, butcontainsFeatureView resources that specify the feature data sources in eitherBigQuery or the Feature Registry. For information about how tocreate an online store instance, seeCreate an online store instance.

• Feature view (FeatureView):AFeatureViewresource is a logical collection of features in an online storeinstance. When you create a feature view, you can specify the location of thefeature data source in either of the following ways:

  • Associate one or more feature groups and features from the Feature Registry. Afeature group specifies the location of the BigQuery data source.A feature within the feature group points to a specific feature columnwithin that data source.

  • Alternatively, associate a BigQuery source table or view.

  For information about how to create feature view instances within an online store,seeCreate a feature view.

For example, figure 3 illustrates a feature view comprising feature columnsf2 andf4, which are sourced from two separate feature groups associatedwith a BigQuery table.

Online serving

Vertex AI Feature Store provides the following types of online servingfor real-time online predictions:

• Bigtable online serving is useful for serving large data volumes(terabytes of data). It's similar to online serving inVertex AI Feature Store (Legacy) and provides improvedcaching. Bigtable online serving doesn't supportembeddings. If youneed to serve large volumes of data that are frequently updated and don't needto serve embeddings, use Bigtable online serving.

• Optimized online serving (Deprecated)lets you online serve features at ultra-low latencies.Optimized online serving also supports embeddings management.

  To use Optimized online serving, you need to configure either a publicendpoint or a dedicated Private Service Connect endpoint.

To learn how to set up online serving in Vertex AI Feature Storeafter you set up features, seeOnline serving types.

Offline serving for batch predictions or model training

Because you don't need to copy or import your feature data from BigQueryto a separate offline store in Vertex AI, you can use the data managementand export capabilities of BigQuery to do the following:

• Query feature data, includinghistorical data at a point in time.

• Preprocess andexport feature data for model training and batch predictions.

For more information about machine learning using BigQuery, seeBigQuery ML introduction.

Vertex AI Feature Store terms

Terms related to feature engineering

feature engineering

• Feature engineering is the process of transforming raw machine learning (ML) data into features that can be used to train ML models or to make inferences.

feature

• In machine learning (ML), a feature is a characteristic or attribute of an instance or entity that's used as an input to train an ML model or to make inferences.

feature value

• A feature value corresponds to the actual and measurable value of a feature (attribute) of an instance or entity. A collection of feature values for the unique entity represent the feature record corresponding to the entity.

feature timestamp

• A feature timestamp indicates when the set of feature values in a specific feature record for an entity were generated.

feature record

• A feature record is an aggregation of all feature values that describe the attributes of a unique entity at a specific point in time.

Terms related to Feature Registry

feature registry

• A feature registry is a central interface for recording feature data sources that you want to serve for online inferences. For more information, seeFeature Registry setup.

feature group

• A feature group is a feature registry resource that corresponds to a BigQuery source table or view containing feature data. A feature view might contain features and can be thought of as a logical grouping of feature columns in the data source.

Terms related to feature serving

feature serving

• Feature serving is the process of exporting or fetching feature values for training or inference. In Vertex AI, there are two types of feature serving—online serving and offline serving. Online serving retrieves the latest feature values of a subset of the feature data source for online inferences. Offline or batch serving exports high volumes of feature data—including historical data—for offline processing, such as ML model training.

offline store

• The offline store is a storage facility storing recent and historical feature data, which is typically used for training ML models. An offline store also contains the latest feature values, which you can serve for online inferences.

online store

• In feature management, an online store is a storage facility for the latest feature values to be served for online inferences.

feature view

• A feature view is a logical collection of features materialized from a BigQuery data source to an online store instance. A feature view stores and periodically refreshes the customer's feature data, which is refreshed periodically from the BigQuery source. A feature view is associated with the feature data storage either directly or through associations to feature registry resources.

Location constraints

All Vertex AI Feature Store resources must be located in the sameregion or the same multi-regional location as your BigQuery datasource. For example, if the feature data source is located inus-central1,you must create yourFeatureOnlineStore instance only inus-central1 or in theUS multi-region location.

Caution: Using source data from dual-region buckets isn't supported.

Feature metadata

Vertex AI Feature Store is integrated with Dataplex Universal Catalog toprovide feature governance capabilities, including feature metadata. Onlinestore instances, feature views, and feature groups are automatically registeredas data assets in Data Catalog, a feature thatcatalogs metadata from these resources. You can then use the metadata searchcapability of Dataplex Universal Catalog to search for, view, and manage themetadata for these resources. For more information about searching forVertex AI Feature Store resources, seeSearch for resource metadata in Data Catalog.

Feature labels

You can add labels to resources during or after the resource creation.For more information about adding labels to existingVertex AI Feature Store resources, seeUpdate labels.

Resource version metadata

Vertex AI Feature Store only supports the version0 for features.

Feature monitoring

Vertex AI Feature Store lets you set up feature monitoring toretrieve feature statistics and detect anomalies in feature data. You can eitherset up monitoring schedules to periodically run monitoring jobs, or manuallyrun a monitoring job. For more information about setting up featuremonitoring and running feature monitoring jobs, seeMonitor features for anomalies.

Embedding management and vector retrieval

Vertex AI Feature Store Optimized online serving isdeprecated. Beginning on May 17, 2026, no new features will be added and only critical patches will be provided. On February 17, 2027, the capability will be fully sunset and APIs will no longer be available.

To improve latency and cost optimizations, migrate toBigtable online serving. To efficiently store and serve embeddings, use the purpose-builtVector Search.

Optimized online serving in Vertex AI Feature Store supportsembedding management. You can store embeddings in BigQuery asregulardouble arrays. Using the embedding management capabilities ofVertex AI Feature Store, you can perform vector similarity searchesto retrieve entities that are approximate nearest neighbors for a specifiedentity or embedding value.

To use embedding management in Vertex AI Feature Store, you need to do the following:

• Set up the BigQuery data source to support embeddings by includingtheembedding column. Optionally, include filtering and crowding columns.For more information, seeData source preparation guidelines.

• Create an online store instance for Optimized online serving.

• Specify theembedding column while creating the feature view. For moreinformation about how to create a feature view that supports embeddings,seeConfigure vector retrieval for a feature view.

For information about how to perform a vector similarity search in Vertex AI Feature Store,seePerform a vector search for entities.

Data retention

Vertex AI Feature Store retains the latest feature values for aunique ID, based on the timestamp associated with the feature values in the datasource. There's no data retention limit in the online store.

Because the offline store is provisioned by BigQuery, dataretention limits or quotas from BigQuery might apply to thefeature data source, including historical feature values.Learn more about quotas and limits in BigQuery.

Quotas and limits

Vertex AI Feature Store enforces quotas and limits to help youmanage resources by setting usage limits, and to protect the community ofGoogle Cloud users by preventing unforeseen spikes in usage. To efficientlyuse Vertex AI Feature Store resources without hitting theseconstraints, review theVertex AI Feature Store quotas and limits.

Pricing

For information about resource usage pricing for Vertex AI Feature Store, seeVertex AI Feature Store pricing.

Notebook tutorials

Use the following samples and tutorials to learn more about Vertex AI Feature Store.

Online feature serving and fetching of BigQuery data with Vertex AI Feature Store Bigtable online serving

In this tutorial, you learn how to use Bigtable online serving in Vertex AI Feature Store for online serving and fetching of feature values in BigQuery. Open in Colab  | Open in Colab Enterprise  | View on GitHub

Online feature serving and fetching of BigQuery data with Vertex AI Feature Store Optimized online serving

In this tutorial, you learn how to use Optimized online serving in Vertex AI Feature Store for serving and fetching of feature values from BigQuery. Open in Colab  | Open in Colab Enterprise  | View on GitHub

Online feature serving and vector retrieval of BigQuery data with Vertex AI Feature Store

In this tutorial, you learn how to use Vertex AI Feature Store for online serving and vector retrieval of feature values in BigQuery. Open in Colab  | Open in Colab Enterprise  | View on GitHub

Vertex AI Feature Store feature view Service Agents

In this tutorial, you learn how to enable feature view Service Agents and grant each feature view access to the specific source data that is used. Open in Colab  | Open in Colab Enterprise  | View on GitHub

Vertex AI Feature Store based LLM grounding tutorial

In this tutorial, you learn how to chunk user-provided data, and then generate embedding vectors for each chunk using a Large Language Model (LLM) that has embedding generation capabilities. The resulting embedding vector dataset can then be loaded into Vertex AI Feature Store, enabling fast feature retrieval and efficient online serving. Open in Colab  | Open in Colab Enterprise  | View on GitHub

Build a GenAI RAG application with Vertex AI Feature Store and BigQuery

In this tutorial, you learn how to build a low-latency vector search system for your Gen AI application using BigQuery vector search and Vertex AI Feature Store. Open in Colab  | Open in Colab Enterprise  | View on GitHub

Configure IAM Policy in Vertex AI Feature Store

In this tutorial, you learn how to configure an IAM policy to control access to resources and data stored within Vertex AI Feature Store. Open in Colab  | Open in Colab Enterprise  | View on GitHub

What's next

• Learn how toset up your data in BigQuery.

• Learn how to createfeature groups andfeatures.

• Learn how tocreate an online store instance.