As previously announced, we're ending support for Inventory data pushed from our Cloud Integrations (AWS, Azure, GCP).
What is changing?
Inventory data is currently collected on many AWS, Azure, and GCP integrations. It represents static configuration data or resource metadata. Information from hosts or on-host integrations in Inventory is not affected by this change.
What is Inventory?
Inventory data is information about the state and configuration of a service, e.g. service settings. Data appears in the Inventory page and some integration dashboards in the original infrastructure experience. Changes to inventory data generate New Relic
InfrastructureEvent events that can be queried and alerted on.
Capabilities that will no longer be available
- When browsing to Infrastructure > Inventory, any source from
gcp/*will no longer be available.
- Built-in dashboards available in Infrastructure > AWS | Azure | GCP | GovCloud that contain inventory or infrastructure events widgets will no longer display data. These dashboards are planned to be migrated as standard New Relic dashboards enabling all the new capabilities (edit, sharing, new visualizations, and more) which are not available in the original dashboard experience for infrastructure monitoring.
- Queries using
InfrastructureEventwill no longer report data.
Why is this capability being deprecated?
- Customers find more value in Host Inventory, which is not being deprecated, but less value in Cloud Integrations.
- Most of the attributes available as Inventory in the polling integrations are also available as metric metadata or tags, so you can query for this information.
- A limited set of integrations also published the same information available in Inventory as standard New Relic Events. In example, the AWS Health integration.
How does this affect AWS CloudWatch metric streams and future integrations for Azure & GCP?
- Inventory isn't available in AWS CloudWatch metric stream, so there's no impact on this capability.
- New integrations for Azure and GCP will be focused on MELT telemetry: Metrics, Events, Logs & Traces, with special focus on Metrics.
If you have any problems after this feature is deprecated, please reach out to us at firstname.lastname@example.org.
- AWS CloudWatch metric stream now supports adding percentile statistics to CloudWatch metrics. The New Relic CloudWatch metric stream integration has been updated in order to automatically ingest percentiles defined in metric streams. Learn more about this capability from the CloudWatch metric stream integration documentation.
- All polling integrations for AWS, Azure, and GCP have been optimized to ensure metrics for existing entities are not impacted when large amount of resources are detected from new or existing provider accounts in a short-period of time. Metrics for new resources will be processed according to New Relic system limits, and may not be ingested until the associated entity is registered.
- AWS CloudWatch metric streams integration has been optimized to ensure resource metrics are always ingested as soon as received. When large amount of new resources are detected in a short-period of time, the associated entity and metadata will be available as soon as the entity is registered.
- A new setting Enable table metrics has been added in the Google BigQuery monitoring integration, disabled by default for new integrations.
- It will be kept enabled for integrations that had the BigQuery integration enabled before the change.
- When disabled BigQuery table entities and their metrics will not be collected.
- A new setting Keep legacy metadata has been added in the AWS EC2 polling integration, disabled by default for new integrations.
- When disabled, legacy attribute formats for EC2 metadata and custom tags will no longer be used. This helps reduce the data ingest quota when those legacy attribute formats are not used in custom dashboards or alert conditions.