New Relic integrations with the Google Cloud Platform (GCP) include one that reports Google Cloud Vertex AI data to New Relic. This document explains how to activate the GCP Vertex AI integration and describes the data it reports.
Features Vertex AI is Google Cloud's unified platform for building, deploying, and managing machine-learning models and generative-AI applications. New Relic Vertex AI integration collects performance, throughput, latency, and resource utilization metrics across locations, endpoints, indexes, feature stores, and feature online stores.
Activate integration To enable the integration, follow standard procedures to connect your GCP service to New Relic :
Polling frequency New Relic integrations query your GCP services according to a polling interval that varies by integration. The polling frequency for Google Cloud Vertex AI is 5 minutes. The resolution is 1 data point every minute.
Workload Identity Federation Find and use data After you enable the integration, your Vertex AI resources appear as entities in the New Relic entity explorer. To see dashboards and manage services, go to one.newrelic.com > All capabilities > Infrastructure > GCP.
All Vertex AI metrics available in GCP Cloud Monitoring are collected as dimensional metrics in the Metric event type. Additional metrics beyond this table are collected automatically. See Google's Vertex AI metrics documentation for the complete list.
Entities Vertex AI entities Entity
Entity type
Resource type
Location
GCPAIPLATFORMLOCATION
aiplatform.googleapis.com/Location
Endpoint
GCPAIPLATFORMENDPOINT
aiplatform.googleapis.com/Endpoint
Index
GCPAIPLATFORMINDEX
aiplatform.googleapis.com/Index
Featurestore
GCPAIPLATFORMFEATURESTORE
aiplatform.googleapis.com/Featurestore
Feature Online Store
GCPAIPLATFORMFEATUREONLINESTORE
aiplatform.googleapis.com/FeatureOnlineStore
Pipeline Job
GCPAIPLATFORMPIPELINEJOB
aiplatform.googleapis.com/PipelineJob
Metric data Key metrics — Location Metric name
Unit
Description
gcp.aiplatform.executing_vertexai_pipeline_jobs
Count
Number of pipeline jobs currently being executed in the location.
gcp.aiplatform.executing_vertexai_pipeline_tasks
Count
Number of pipeline tasks currently being executed in the location.
gcp.aiplatform.online_prediction_requests_per_base_model
Count
Number of online prediction requests per project per base model.
gcp.aiplatform.quota.online_prediction_requests_per_base_model.usage
Count
Current quota usage for online prediction requests per base model.
gcp.aiplatform.quota.online_prediction_requests_per_base_model.limit
Count
Current quota limit for online prediction requests per base model.
gcp.aiplatform.quota.online_prediction_requests_per_base_model.exceeded
Count
Number of attempts that exceeded the online prediction requests quota.
gcp.aiplatform.online_prediction_input_tokens_per_minute_per_base_model
Count
Online prediction input tokens per minute per project per base model.
gcp.aiplatform.online_prediction_output_tokens_per_minute_per_base_model
Count
Online prediction output tokens per minute per project per base model.
gcp.aiplatform.publisher.online_serving.token_count
Count
Total tokens served by publisher models in the location.
gcp.aiplatform.publisher.online_serving.first_token_latencies
Milliseconds
Latency to first response token from publisher models.
For the complete list of location metrics, see Google's Vertex AI metrics documentation .
Key metrics — Endpoint Metric name
Unit
Description
gcp.aiplatform.prediction.online.accelerator.duty_cycle
Percent
Average fraction of time over the past sample period during which the accelerators were actively processing.
gcp.aiplatform.prediction.online.accelerator.memory.bytes_used
Bytes
Accelerator memory allocated by the deployed model replica.
gcp.aiplatform.prediction.online.cpu.utilization
Percent
CPU utilization of the deployed model replica.
gcp.aiplatform.prediction.online.memory.bytes_used
Bytes
Memory in use by the deployed model replica.
gcp.aiplatform.prediction.online.network.received_bytes_count
Bytes
Network bytes received by the deployed model replica.
gcp.aiplatform.prediction.online.network.sent_bytes_count
Bytes
Network bytes sent by the deployed model replica.
gcp.aiplatform.prediction.online.prediction_count
Count
Number of online predictions served by the endpoint.
gcp.aiplatform.prediction.online.prediction_latencies
Milliseconds
Online prediction latency of the deployed model.
gcp.aiplatform.prediction.online.error_count
Count
Number of online prediction errors returned by the endpoint.
gcp.aiplatform.prediction.online.response_count
Count
Number of online prediction responses returned by the endpoint, faceted by response code.
gcp.aiplatform.prediction.online.replicas
Count
Number of active replicas serving the deployed model.
gcp.aiplatform.prediction.online.target_replicas
Count
Target number of active replicas for the deployed model.
For the complete list of endpoint metrics, see Google's Vertex AI metrics documentation .
Key metrics — Index Metric name
Unit
Description
gcp.aiplatform.matching_engine.current_shards
Count
Current number of shards backing the index.
gcp.aiplatform.matching_engine.stream_update.request_count
Count
Number of stream-update requests sent to the index.
gcp.aiplatform.matching_engine.stream_update.datapoint_count
Count
Number of datapoints successfully upserted or removed via stream update.
gcp.aiplatform.matching_engine.stream_update.latencies
Milliseconds
Latency between a stream-update response and the update taking effect.
For the complete list of index metrics, see Google's Vertex AI metrics documentation .
Key metrics — Featurestore Metric name
Unit
Description
gcp.aiplatform.featurestore.cpu_load
Percent
Average CPU load across nodes in the Featurestore online storage.
gcp.aiplatform.featurestore.cpu_load_hottest_node
Percent
CPU load on the hottest node in the Featurestore online storage.
gcp.aiplatform.featurestore.node_count
Count
Number of nodes provisioned for the Featurestore online storage.
gcp.aiplatform.featurestore.online_entities_updated
Count
Number of entities updated in the Featurestore online storage.
gcp.aiplatform.featurestore.online_serving.request_count
Count
Number of online serving requests handled by the Featurestore, faceted by EntityType.
gcp.aiplatform.featurestore.online_serving.latencies
Milliseconds
Online serving request latency at the Featurestore, faceted by EntityType.
gcp.aiplatform.featurestore.online_serving.request_bytes_count
Bytes
Online serving request size at the Featurestore, faceted by EntityType.
gcp.aiplatform.featurestore.online_serving.response_size
Bytes
Online serving response size at the Featurestore, faceted by EntityType.
gcp.aiplatform.featurestore.storage.stored_bytes
Bytes
Total data stored in the Featurestore.
gcp.aiplatform.featurestore.storage.billable_processed_bytes
Bytes
Billable bytes processed for Featurestore offline data.
gcp.aiplatform.featurestore.streaming_write.offline_processed_count
Count
Number of streaming-write requests processed into offline storage.
gcp.aiplatform.featurestore.streaming_write.offline_write_delays
Seconds
Time from when the streaming-write API is called until the record reaches offline storage.
For the complete list of featurestore metrics, see Google's Vertex AI metrics documentation .
Key metrics — Feature Online Store Metric name
Unit
Description
gcp.aiplatform.featureonlinestore.online_serving.request_count
Count
Number of online serving requests handled by the Feature Online Store, faceted by FeatureView.
gcp.aiplatform.featureonlinestore.online_serving.serving_latencies
Milliseconds
Online serving request latency at the Feature Online Store, faceted by FeatureView.
gcp.aiplatform.featureonlinestore.online_serving.serving_bytes_count
Bytes
Online serving response size at the Feature Online Store, faceted by FeatureView.
gcp.aiplatform.featureonlinestore.running_sync
Count
Number of syncs currently running at the Feature Online Store.
gcp.aiplatform.featureonlinestore.serving_data_ages
Seconds
Age of the data being served by the Feature Online Store.
gcp.aiplatform.featureonlinestore.serving_data_by_sync_time
Count
Breakdown of records in the Feature Online Store by synced timestamp.
gcp.aiplatform.featureonlinestore.storage.bigtable_cpu_load
Percent
Average CPU load across Bigtable nodes backing the Feature Online Store.
gcp.aiplatform.featureonlinestore.storage.bigtable_cpu_load_hottest_node
Percent
CPU load on the hottest Bigtable node backing the Feature Online Store.
gcp.aiplatform.featureonlinestore.storage.bigtable_nodes
Count
Number of Bigtable nodes backing the Feature Online Store.
gcp.aiplatform.featureonlinestore.storage.stored_bytes
Bytes
Total data stored in the Feature Online Store.
For the complete list of feature online store metrics, see Google's Vertex AI metrics documentation .
Key metrics — Pipeline Job Metric name
Unit
Description
gcp.aiplatform.pipelinejob.duration
Seconds
Runtime seconds of the pipeline job being executed, from creation to end.
gcp.aiplatform.pipelinejob.task_completed_count
Count
Total number of completed pipeline tasks in the pipeline job.
For the complete list of pipeline job metrics, see Google's Vertex AI metrics documentation .
Metrics-only resource types New Relic also collects metrics for the following Vertex AI resource types. These resource types don't create entities in the entity explorer, so query their metrics directly on the Metric event type.
Publisher Model — 15 metrics Metric name
Unit
Description
gcp.aiplatform.publisher.online_serving.character_count
Count
Accumulated input/output character count.
gcp.aiplatform.publisher.online_serving.characters
Count
Input/output character count distribution.
gcp.aiplatform.publisher.online_serving.consumed_throughput
Count
Overall throughput used (accounting for burndown rate) in terms of characters.
gcp.aiplatform.publisher.online_serving.consumed_token_throughput
Count
Overall throughput used (accounting for burndown rate) in terms of tokens.
gcp.aiplatform.publisher.online_serving.dedicated_character_limit
Per second
Dedicated limit in characters per second.
gcp.aiplatform.publisher.online_serving.dedicated_character_project_max_limit
Per second
Project max limit in characters per second that can be consumed.
gcp.aiplatform.publisher.online_serving.dedicated_gsu_limit
Count
Dedicated limit in GSU.
gcp.aiplatform.publisher.online_serving.dedicated_gsu_project_max_limit
Count
Project max limit in GSU that can be consumed.
gcp.aiplatform.publisher.online_serving.dedicated_token_limit
Per second
Dedicated limit in tokens per second.
gcp.aiplatform.publisher.online_serving.dedicated_token_project_max_limit
Per second
Project max limit in tokens per second that can be consumed.
gcp.aiplatform.publisher.online_serving.first_token_latencies
Milliseconds
Duration from request received to first token sent back to the client.
gcp.aiplatform.publisher.online_serving.model_invocation_count
Count
Number of model invocations (prediction requests).
gcp.aiplatform.publisher.online_serving.model_invocation_latencies
Milliseconds
Model invocation latencies (prediction latencies).
gcp.aiplatform.publisher.online_serving.token_count
Count
Accumulated input/output token count.
gcp.aiplatform.publisher.online_serving.tokens
Count
Input/output token count distribution.
Deployment Resource Pool — 8 metrics Metric name
Unit
Description
gcp.aiplatform.prediction.online.deployment_resource_pool.accelerator.duty_cycle
Percent
Average fraction of time over the past sample period during which the accelerator(s) were actively processing.
gcp.aiplatform.prediction.online.deployment_resource_pool.accelerator.memory.bytes_used
Bytes
Amount of accelerator memory allocated by the deployment resource pool replica.
gcp.aiplatform.prediction.online.deployment_resource_pool.cpu.utilization
Percent
Fraction of CPU allocated by the deployment resource pool replica and currently in use. May exceed 100% if the machine type has multiple CPUs.
gcp.aiplatform.prediction.online.deployment_resource_pool.memory.bytes_used
Bytes
Amount of memory allocated by the deployment resource pool replica and currently in use.
gcp.aiplatform.prediction.online.deployment_resource_pool.network.received_bytes_count
Bytes
Number of bytes received over the network by the deployment resource pool replica.
gcp.aiplatform.prediction.online.deployment_resource_pool.network.sent_bytes_count
Bytes
Number of bytes sent over the network by the deployment resource pool replica.
gcp.aiplatform.prediction.online.deployment_resource_pool.replicas
Count
Number of active replicas used by the deployment resource pool.
gcp.aiplatform.prediction.online.deployment_resource_pool.target_replicas
Count
Target number of active replicas needed for the deployment resource pool.
Reasoning Engine — 8 metrics Metric name
Unit
Description
gcp.aiplatform.reasoning_engine.cpu.allocation_time
CPU-seconds
Container CPU allocation in seconds.
gcp.aiplatform.reasoning_engine.memory.allocation_time
GiB-seconds
Container memory allocation in Gigabytes-seconds.
gcp.aiplatform.reasoning_engine.memory_bank.generate_memories_token_count
Count
The number of tokens used in memory generation requests per Reasoning Engine.
gcp.aiplatform.reasoning_engine.memory_bank.memory_lro_latency
Milliseconds
The total latency of memory LROs.
gcp.aiplatform.reasoning_engine.memory_bank.memory_mutation_count
Count
The number of memory mutation operations on the Reasoning Engine.
gcp.aiplatform.reasoning_engine.memory_bank.memory_retrieval_count
Count
The number of memory entities retrieved per Reasoning Engine.
gcp.aiplatform.reasoning_engine.request_count
Count
Number of requests reaching the instance.
gcp.aiplatform.reasoning_engine.request_latencies
Milliseconds
Distribution of request latency in milliseconds reaching the instance.
Index Endpoint — 6 metrics Metric name
Unit
Description
gcp.aiplatform.matching_engine.current_replicas
Count
Number of active replicas used by the DeployedIndex.
gcp.aiplatform.matching_engine.current_shards
Count
Number of shards of the DeployedIndex.
gcp.aiplatform.matching_engine.query.latencies
Milliseconds
Server side request latency.
gcp.aiplatform.matching_engine.query.request_count
Count
Number of requests received.
gcp.aiplatform.matching_engine.cpu.request_utilization
Percent
The fraction of the requested CPU that is currently in use on a match server container.
gcp.aiplatform.matching_engine.memory.used_bytes
Bytes
The memory used in bytes for a match server container.
Model — 3 metrics Metric name
Unit
Description
gcp.aiplatform.model_monitoring.feature_attribution_deviation
Count
The feature attribution deviation value of the given feature.
gcp.aiplatform.model_monitoring.feature_drift_deviation
Count
The drift deviation value of the given feature.
gcp.aiplatform.model_monitoring.prediction_output_drift_deviation
Count
The drift deviation value of the prediction output.
Online Evaluator — 2 metrics Metric name
Unit
Description
gcp.aiplatform.online_evaluator.evaluations_count
Count
The number of evaluations of the Online Evaluator resource.
gcp.aiplatform.online_evaluator.scores
Count
The distribution of scores of the Online Evaluator.
Service account or user account Find and use data After activating the integration and waiting a few minutes (based on the polling frequency ), data will appear in the New Relic UI. To find and use your data , including links to your dashboards and alert settings, go to one.newrelic.com > All capabilities > Infrastructure > GCP > (select an integration).
Data is attached to the following event types :
Entity
Event Type
Provider
Endpoint
GcpVertexAiEndpointSample
GcpVertexAiEndpoint
Feature store
GcpVertexAiFeaturestoreSample
GcpVertexAiFeaturestore
Feature Online Store
GcpVertexAiFeatureOnlineStoreSample
GcpVertexAiFeatureOnlineStore
Location
GcpVertexAiLocationSample
GcpVertexAiLocation
Index
GcpVertexAiIndexSample
GcpVertexAiIndex
PipelineJob
GcpVertexAiPipelineJobSample
GcpVertexAiPipelineJob
For more on how to use your data, see Understand and use integration data .
VertexAI Endpoint data Metric
Unit
Description
prediction.online.accelerator.duty_cycle
Percent
Average fraction of time over the past sample period during which the accelerator(s) were actively processing.
prediction.online.accelerator.memory.bytes_used
Bytes
Amount of accelerator memory allocated by the deployed model replica.
prediction.online.error_count
Count
Number of online prediction errors.
prediction.online.memory.bytes_used
Bytes
Amount of memory allocated by the deployed model replica and currently in use.
prediction.online.network.received_bytes_count
Bytes
Number of bytes received over the network by the deployed model replica.
prediction.online.network.sent_bytes_count
Bytes
Number of bytes sent over the network by the deployed model replica.
prediction.online.prediction_count
Count
Number of online predictions.
prediction.online.prediction_latencies
Milliseconds
Online prediction latency of the deployed model.
prediction.online.private.prediction_latencies
Milliseconds
Online prediction latency of the private deployed model.
prediction.online.replicas
Count
Number of active replicas used by the deployed model.
prediction.online.response_count
Count
Number of different online prediction response codes.
prediction.online.target_replicas
Count
Target number of active replicas needed for the deployed model.
VertexAI Featurestore data Metric
Unit
Description
featurestore.cpu_load
Percent
The average CPU load for a node in the Featurestore online storage.
featurestore.cpu_load_hottest_node
Percent
The CPU load for the hottest node in the Featurestore online storage.
featurestore.node_count
Count
The number of nodes for the Featurestore online storage.
featurestore.online_entities_updated
Count
Number of entities updated on the Featurestore online storage.
featurestore.online_serving.latencies
Milliseconds
Online serving latencies by EntityType.
featurestore.online_serving.request_bytes_count
Bytes
Request size by EntityType.
featurestore.online_serving.request_count
Count
Featurestore online serving count by EntityType.
featurestore.online_serving.response_size
Bytes
Response size by EntityType.
featurestore.storage.billable_processed_bytes
Bytes
Number of bytes billed for offline data processed.
featurestore.storage.stored_bytes
Bytes
Bytes stored in Featurestore.
featurestore.streaming_write.offline_processed_count
Count
Number of streaming write requests processed for offline storage.
featurestore.streaming_write.offline_write_delays
Seconds
Time (in second) since the write API is called until it is written to offline storage.
VertexAI FeatureOnlineStore data Metric
Unit
Description
featureonlinestore.online_serving.request_count
Count
Number of serving count by FeatureView.
featureonlinestore.online_serving.serving_bytes_count
Bytes
Serving response size by FeatureView.
featureonlinestore.online_serving.serving_latencies
Milliseconds
Online serving latencies by FeatureView.
featureonlinestore.running_sync
Milliseconds
Number of running syncs at given point of time.
featureonlinestore.serving_data_ages
Seconds
Measure of the serving data age in seconds.
featureonlinestore.serving_data_by_sync_time
Count
Breakdown of data in Feature Online Store by synced timestamp.
featureonlinestore.storage.bigtable_cpu_load
Percent
The average CPU load of nodes in the Feature Online Store.
featureonlinestore.storage.bigtable_cpu_load_hottest_node
Percent
The CPU load of the hottest node in the Feature Online Store.
featureonlinestore.storage.bigtable_nodes
Count
The number of nodes for the Feature Online Store(Bigtable).
featureonlinestore.storage.stored_bytes
Count
Bytes stored in the Feature Online Store.
VertexAI Location data Metric
Unit
Description
online_prediction_requests_per_base_model
Count
Number of requests per base model.
quota.online_prediction_requests_per_base_model.exceeded
Count
Number of attempts to exceed the limit on quota metric.
quota.online_prediction_requests_per_base_model.limit
Count
Current limit on quota metric.
quota.online_prediction_requests_per_base_model.usage
Count
Current usage on quota metric.
executing_vertexai_pipeline_jobs
Count
Number of pipeline jobs being executed.
executing_vertexai_pipeline_tasks
Count
Number of pipeline tasks being executed.
VertexAI Index data Metric
Unit
Description
matching_engine.stream_update.datapoint_count
Count
Number of successfully upserted or removed datapoints.
matching_engine.stream_update.latencies
Milliseconds
The latencies between the user receives a UpsertDatapointsResponse or RemoveDatapointsResponse and that update takes effect.
matching_engine.stream_update.request_count
Count
Number of stream update requests.
VertexAI Pipeline Job data Metric
Unit
Description
pipelinejob.duration
Seconds
Runtime seconds of the pipeline job being executed (from creation to end).
pipelinejob/task_completed_count
Count
Total number of completed Pipeline Tasks.