Prepare and test your skills

Prepare and test your skills

Common API types and interaction models in Google Cloud are different styles for calling the same kinds of platform services. Google Distributed Cloud (GDC) (opens in a new tab) services expose REST, RPC, and Model Context Protocol (MCP) (Preview) interfaces, and applications reach them through direct calls, client libraries, gcloud CLI commands, and MCP tools or servers. The choice of style changes how the client connects, while the underlying capability such as computing, networking, storage, or machine-learning-based data analysis stays the same.
REST, RPC, and MCP are three API interaction styles that differ in how a client connects to a Google Cloud service. Service Health (opens in a new tab) shows the overlap clearly because it offers both a REST API and an RPC API for listing health events and getting event details. REST clients use REST interfaces, RPC clients use RPC client stubs, and MCP clients use MCP tools or connect to an MCP server.
Three side-by-side cards compare REST, RPC, and MCP Preview by how clients connect and example services, showing Service Health supports both REST and RPC for the same capability.
| API style | Example services | How clients interact |
|---|---|---|
| REST | Service Health REST API, Cloud Build (opens in a new tab) REST API, Cloud Storage JSON API | REST interfaces, client libraries, gcloud CLI |
| RPC | Service Health RPC API, App Design Center RPC APIs | RPC client stubs |
| MCP | Service Health MCP tools, App Design Center MCP server | MCP tools or MCP server connection |
The Cloud Storage JSON API uses JSON for its REST interface, and MCP is Preview. REST and RPC can expose the same capability with different client interaction, while MCP is a separate preview model for tool-based connections.
Exam tip: REST APIs are called through REST interfaces, RPC APIs through RPC client stubs, and MCP through MCP tools or an MCP server.
Client libraries and the gcloud CLI are two practical paths that sit between the user and a service API. Client libraries (opens in a new tab) are programming-language-specific packages that make it easier to call service APIs, and Service Health and Cloud Build both provide them. The Google API Client Library for Python is an older library for easily accessing REST interfaces. The gcloud CLI is a command-line path where users run gcloud commands to manage resources or request events for a project or organization, and the CLI then calls the service API on their behalf.
Google Distributed Cloud air-gapped APIs (opens in a new tab) are programmatic interfaces to GDC platform services that fall into three groups with different purposes. GDC Service APIs let applications add computing, networking, storage, and machine-learning-based data analysis functionality. System APIs are used by the underlying GDC system itself, are discoverable in Kubernetes with kubectl and other open source tools, and are restricted to the relevant Infrastructure Operators. Kubernetes APIs are the third listed group, and the discovery flow runs from operator tools such as kubectl to the Kubernetes API surface. This creates a clear boundary where Service APIs are for applications and System APIs are reserved for operators.
The core purpose of an application programming interface (API) is to let one application request data or functionality from another service in a structured way. In Google Cloud, APIs expose platform services so applications can add computing, networking, storage, and machine-learning-based data analysis without rebuilding those services. Each API defines how requests are made and how responses are returned, so services, applications, and systems can communicate reliably and securely.
APIs in Google Cloud act as standardized interfaces grouped by the kind of capability they add. This grouping helps choose an API based on the function needed rather than building a custom connection for each application.
Access boundaries in Google Distributed Cloud (GDC) (opens in a new tab) separate APIs meant for applications from APIs reserved for the platform itself. GDC Service APIs are programmatic interfaces that applications call to add platform functionality such as computing, networking, storage, and machine-learning-based data analysis. System APIs are used by the underlying GDC system, are discoverable in Kubernetes, and are restricted to the relevant Infrastructure Operators. Kubernetes APIs are also part of GDC, and kubectl and other open source tools can discover System APIs in Kubernetes. The result is that regular applications call Service APIs, while only Infrastructure Operators operate on the System side of the boundary.
Exam tip: Service APIs are for applications to use platform services, while System APIs are used by the underlying GDC system and are restricted to Infrastructure Operators.
Reliable and secure API communication depends on documented methods, identity checks, helper libraries, and access controls working together. Application Integration (opens in a new tab) requires authentication when using REST, which proves the caller's identity before access is granted. API references document the available methods and how to call them, such as the Cloud Build REST overview and the Application Integration V1 and V2 references. Client libraries (opens in a new tab) make REST interfaces easier to call from code, and Cloud Build and Service Health both provide them. Access controls limit who can use sensitive surfaces, such as GDC System APIs restricted to Infrastructure Operators. Multiple access methods give developers choice, including REST APIs, RPC APIs, the gcloud CLI (opens in a new tab), client libraries, and Preview MCP servers, while schemas such as the Service Health event log schema define the data format for viewing, exporting, or storing events.
The value of APIs in cloud-based integration is that they expose reusable capabilities that many applications can call instead of rebuilding them. An API in Google Cloud is a programmatic connection point where the call flows from the application to the service that owns the capability, and the service returns the result to the application. Because the API defines the request and response, separately built services, applications, and systems can interoperate, which supports modular design, automation, and innovation.
Reusable capabilities are platform functions that applications add by calling Google Distributed Cloud (GDC) (opens in a new tab) air-gapped APIs without managing the underlying platform. Applications use these Service APIs to add computing functionality to run workloads, networking functionality to connect systems, storage functionality to save and retrieve data, and machine-learning-based data analysis functionality to gain insights from data. The application depends on the API rather than the internal implementation, so the same service can be reused across many applications. This reuse removes the need for custom connections between every pair of systems.
Interoperability comes from combining separate Service APIs so data flows from one service to the next through API requests. A single workflow can call a compute API to run code, a storage API to save results, and a data API to analyze them, with each service handling only its bounded job. Common groups that compose this way include Compute APIs such as Compute Engine API and Kubernetes Engine API, Storage and Database APIs such as Cloud Storage JSON API, Cloud SQL Administration API, and Cloud Spanner API, Networking APIs such as Cloud DNS API and Service Directory API, Big Data APIs such as BigQuery API, Cloud Dataflow API, and Cloud Pub/Sub API, Machine Learning APIs such as Cloud Vision API, Cloud Speech API, and Cloud Translation API, and Management Tool APIs such as Cloud Build API (opens in a new tab) and Deployment Manager API. Together these groups let developers reuse the same cloud services instead of creating one-off integrations.
Modular design means building an application from separate pieces that each handle one job, and APIs make that separation work by letting modules compose through calls. The management chain shows this composition where the App Hub API organizes existing resources into applications, the Application Design Center API (opens in a new tab) designs and deploys applications, the Infrastructure Manager API automates deployment of infrastructure resources, and the Cloud Build API runs builds for those deployments. These APIs also have dependencies between them, where Pub/Sub supports Cloud Build and Secret Manager supports Developer Connect, so enabling one capability can require its supporting API. Teams can therefore add, replace, or scale one module without rebuilding the whole application because each module depends on stable API contracts.
Controlled boundaries in GDC separate open innovation on Service APIs from protected platform operations on System APIs. GDC Service APIs are the interfaces that applications call to use platform services, while System APIs are used by the underlying GDC system itself. The discovery flow runs from operator tools such as kubectl and other open source tools to the Kubernetes API surface where System APIs are discoverable. Access to System APIs sits on the operator side of the boundary and is restricted to the relevant Infrastructure Operators, so regular applications cannot call them directly. This separation protects platform operations while still letting applications build new solutions on top of Service APIs.
Exam tip: GDC Service APIs are for applications adding platform functionality, while System APIs are discoverable with kubectl but restricted to Infrastructure Operators.
An application programming interface (API) lets one application request data or functionality from another service in a structured way. In Google Cloud, APIs expose platform services so applications can add computing, networking, storage, and machine-learning-based data analysis without rebuilding those services, and each API defines how requests are made and how responses are returned.
REST, RPC, and Model Context Protocol (MCP) differ in how a client connects to a Google Cloud service. REST clients use REST interfaces, RPC clients use RPC client stubs, and MCP clients use MCP tools or an MCP server connection, with MCP in Preview. REST and RPC can expose the same capability with different client interaction, while MCP is a separate preview model for tool-based connections.
Google Distributed Cloud Service APIs are programmatic interfaces that applications call to add platform functionality such as computing, networking, storage, and machine-learning-based data analysis. System APIs are used by the underlying Google Distributed Cloud system itself, are discoverable in Kubernetes with kubectl and other open source tools, and are restricted to the relevant Infrastructure Operators.
An engineering team is modernizing an internal microservices application running on Google Cloud. The microservices need to exchange high volumes of real-time data with minimal latency, support bidirectional streaming, and serialize payloads efficiently using compact binary data formats.
Which API style or framework should the team adopt for inter-service communication?
GraphQL using JSON queries
REST using JSON over HTTP/1.1
SOAP using XML over HTTP
gRPC using Protocol Buffers
Last updated