Datadog API Catalog

Join the Beta!

API Catalog is in limited availability. Use this form to get early access.

API Catalog showing a list of API endpoints along with their Error Rate, P99 Latency, Requests Count, and Owning Team, with searching, filtering, and scoping features.

Overview

The API Catalog provides a single view and entry point for exploring the performance, reliability, and ownership of all your API endpoints in one place. It’s a central place where your whole company can find up-to-date information about the characteristics of the APIs used by internal services (private APIs) and external users (publicly exposed APIs).

Monitor your mission-critical API-driven business features, standardize and validate APIs performance expectations, and alert when performance deviates from them.

The API Catalog enables your teams to:

Provide high availability and uptime of critical APIs and the business features that rely on them to function.
Prevent API regressions and instability issues.
Quickly triage incidents.

API Catalog combines data from across Datadog to provide opinionated workflows so you can explore and monitor your APIs from different sources in one unified view. API Catalog provides:

Automated discoverability - One inventory for all public, private, and partner APIs, where endpoints are the organizing principle.
Correlation of all API metadata from different Datadog sources.
API endpoint metrics, such as Last Seen, Requests, Latency, and Errors, to identify performance issues and track API health.
Alerting on endpoints that deviate from defined performance expectations and thresholds.
Resolving incidents faster with API ownership information (team, on-call, communication channels) directly associated with each endpoint, to know who to reach when something goes wrong

See Key terminology for some background on concepts that are important to understanding what API Catalog does.

Exploring your endpoints

The API Catalog view shows all endpoints in all environments in your Datadog organization. The performance data shown for each endpoint is scoped to the environment and time frame you select. You can browse and ask questions by querying different properties and metrics to get more precise results, using facets and tags for quick filtering.

In the list table, sort by any of the columns by clicking a header. For example, click P99 to see endpoints with the highest 99th percentile for latency.

The table also shows team ownership for the API. This information is inherited from the associated service definition in the Service Catalog. The service owner owns all of the endpoints connected to the service.

To filter the list of endpoints or to search for a particular endpoint of interest, enter a query in the Search field. Search by service, path, or any other primary tag.

Or select a combination of facets on the left, filtering by the owning service or the team they belong to.

To scope the data shown in the table, specify an environment, another primary tag (such as datacenter), and a time frame.

You can define tags to use as facets so you can easily find groups of endpoints you’re most interested in.

Investigating endpoint details

When exploring an endpoint, getting a higher resolution view of the API endpoint details is helpful.

Click an endpoint in the list to open a details page that shows performance, ownership, and metadata information for the endpoint, collected from various areas of Datadog, into one place. Here you can edit the short name and description and add custom tags. You can also launch deeper investigations of the various types of telemetry using links into other areas of Datadog.

For example, you can:

Look up a specific endpoint by path (for example, /checkout) when it shows a high error rate and a high request count.
View the Error/Hits graph and correlated Response Code graph to identify issues.
Navigate to related telemetry like traces and logs.

The performance graphs on the page are initially scoped to the same settings as on the API Catalog page, and you can change those settings on the details page to suit your investigation by using the time frame selector and other scope dropdown menus.

Exploring endpoints during incidents

When investigating an incident, getting a detailed resolution view of the API endpoint can lead to a faster understanding of the root cause.

In addition, you can quickly identify the owning team of each endpoint, the on-call engineer, and how to reach them (email, Slack, PagerDuty) to help resolve incidents involving a specific API endpoint.

The ownership information – team, on-call information, communication details – is derived from the service definition supplied from the Service Catalog.

The team details panel in the endpoint details page, showing the name and communication information for the team that owns the endpoint, as defined in Service Catalog

Group APIs to express a feature or business logic

In addition to tagging an endpoint in its details page, you can group endpoints by adding tags to multiple endpoints at once. Select multiple endpoint check boxes, and click Edit tags to provide labels that describe business logic, importance, or other useful grouping information. Applying these labels can help you quickly view and access groups of endpoints defined by your own criteria, and create assets such as monitors and dashboards for endpoint groups with common ground and expectations.

For example, if you want to create latency alerts for endpoints that are particularly sensitive to performance problems, tag those endpoints with a tag like Latency sensitive.

Or tag endpoints that handle sensitive data with a tag like PII and alert when endpoints with that tag have a 401 Unauthorized response. Other examples of useful tags:

critical to feature X
newly added - V2
contains password
contains PII
Business logic (for example, payments )
Specific customer group (subscribers)

When you add a tag, it appears in the list of facets on the left of the catalog. Click a facet to filter the list and add the tag to the Search query field.

Setting up

Automatically built on APM data, API Catalog can be turned on in an existing APM instance with no other setup required. Only APM instrumented services and supported libraries are discoverable.

To set up your API Catalog list:

Instrument the services with APM.
Define service ownership information using Service Catalog.

Key terminology

API: A set of protocols and tools that allow two applications to communicate.
API endpoint: The address of a resource (URL) of a server or a service that implements the set of rules defined in the API, often through an HTTP, RESTful API interface. The API endpoint responsible for making the API call response.

The API Catalog displays API endpoints as: the HTTP method (for example, GET), the URL path (the structure of the location of the resource, such as /payment/{shop_id}/purchase), and the name of the service this resource serves (for example, Payments)

The API Catalog in beta supports only HTTP endpoints.
Public APIs: Customer-facing API endpoints that are accessible from the internet.
Private APIs: Also called internal APIs. Intended solely for internal use within a company or organization, used mainly for backend service communication. The most common type of APIs.
Partner APIs: Also called third-party APIs. Another organization’s public endpoints that your organization uses to provide your services (for example, Stripe, Google, and Facebook).