GCP Professional Cloud Developer key terms

An application pattern where each request carries all the information needed to process it and no session data is stored on the instance. Any replica can therefore handle any request, so the service scales horizontally and survives instance loss.

A property of an operation that produces the same result whether it runs once or many times. It lets clients safely retry after timeouts without creating duplicate records or charging a customer twice.

A retry strategy where the wait between attempts grows exponentially, with a random offset added so many clients do not retry at the same instant. It reduces thundering-herd load on a recovering service.

A resilience pattern that stops sending requests to a failing dependency once an error threshold is crossed, failing fast instead. After a cool-down period it probes again and resumes traffic when the dependency recovers.

A periodic check telling a container orchestrator whether an instance is prepared to receive traffic. Instances that fail it are pulled from load balancing until they pass, preventing requests from reaching a still-starting process.

The extra latency a request experiences when it reaches a serverless instance with no warm runtime yet, forcing the platform to allocate resources and initialize the code before the handler executes.

Adding or removing identical instances in response to demand, driven by metrics such as CPU utilization, request concurrency, or a Pub/Sub backlog. It differs from resizing a single machine to be larger.

Running the same application in two or more geographic regions so a regional outage does not take the service offline and users are served from a nearby location. It requires data replication and traffic routing decisions.

Controls that cap how many requests or resources a project, user, or API key may consume in a given period. They protect backends from overload and stop one tenant from starving the others.

A non-human identity used by code and virtual machines to call Google Cloud APIs. Permissions come from IAM roles bound to the account, and applications authenticate as it instead of using a person's credentials.

A way to let workloads running outside Google Cloud, such as those on AWS or GitHub Actions, obtain short-lived Google Cloud credentials by exchanging an external identity token, avoiding downloaded service account keys.

A hosted store for API keys, passwords, and certificates that supports versioning, audit logging, and IAM-controlled access. Applications fetch secrets at runtime rather than embedding them in code or container images.

A security perimeter around Google Cloud services that blocks data movement to resources outside the perimeter, helping prevent data exfiltration even when credentials are compromised.

A managed resource that lets serverless workloads such as Cloud Run or Cloud Functions reach private IP addresses in a VPC network, for example to call a database or an internal load balancer.

A URL granting temporary, time-limited access to a specific Cloud Storage object. It carries a cryptographic signature proving the issuer has permission, so the recipient needs no Google account.

A serverless container platform that runs stateless HTTP services, scaling instances from zero to many based on incoming requests and billing only for resources used while handling them.

A managed Kubernetes service where Google operates the control plane while node pools run containerized workloads, offering autoscaling, Autopilot mode, and integration with IAM and VPC networking.

A client library behavior that searches the environment for credentials in a defined order, such as a local gcloud login or an attached service account, so code needs no hardcoded keys.

A managed continuous integration and delivery service that runs build steps defined in a configuration file, producing container images or artifacts. Triggers can start builds from repository pushes or pull requests.

An immutable, layered package containing an application and its dependencies, built from a Dockerfile. Registries store images and orchestrators pull them by digest or tag to start containers.

A local stand-in for a Google Cloud service, such as Pub/Sub or Firestore, that lets tests run offline and deterministically without creating billable cloud resources.

Writing log entries as machine-readable JSON with named fields instead of free text. The payload becomes searchable in Cloud Logging, and well-known fields such as severity and trace ID are indexed automatically.

A distributed tracing service that records the timing of requests across services and shows where latency accumulates. It accepts spans from instrumented applications and integrates with OpenTelemetry.

A service that aggregates and counts application exceptions and crashes from logs and stack traces, grouping similar errors together and alerting when a new error type appears.

A managed repository for container images and language packages such as Maven or npm artifacts, with IAM-based access control and vulnerability scanning. It replaces Container Registry for new projects.

A managed delivery service for releasing container images to GKE, Cloud Run, or Anthos through ordered environments, supporting approval gates and automated promotion between deployment targets.

An infrastructure-as-code tool that declares Google Cloud resources in configuration files and applies them through providers, keeping a state file to track which resources already exist.

Releasing a new version to a small fraction of traffic first, watching error rates and latency, then gradually increasing the share or rolling back if the signals degrade.

Maintaining two identical environments where traffic switches from the current version to the new one all at once, allowing an instant rollback by switching traffic back.

Directing a defined percentage of requests to different revisions or versions of a service, used for gradual rollouts and A/B comparisons. Both Cloud Run and GKE ingress support it.

A durable, asynchronous messaging service with topics and subscriptions that decouples publishers from subscribers and buffers spikes, supporting at-least-once delivery with push or pull consumers.

A separate Pub/Sub topic where messages that a subscription fails to acknowledge after a configured number of delivery attempts are forwarded, so they can be inspected without blocking the pipeline.

A service that routes events from Google Cloud sources, such as Cloud Storage changes or audit logs, to Cloud Run, Cloud Functions, or GKE targets using the CloudEvents format and filters.

A service for queueing asynchronous work items with rate and concurrency limits and scheduled dispatch to HTTP endpoints, useful for jobs that must be retried with controlled pacing.

A managed cron service that invokes HTTP endpoints, publishes to Pub/Sub topics, or calls App Engine handlers on a schedule, with retry configuration and time zone support.

A managed front door for APIs that handles authentication, API keys, quotas, and routing to backends described by an OpenAPI specification, without deploying additional infrastructure.

A serverless document database with real-time listeners, offline SDK support, and strong consistency per document, suited to mobile and web application data that must scale automatically.

A globally distributed relational database that offers strong consistency and horizontal scale through replicated regional or multi-region configurations, queried with SQL and client libraries.

A serverless analytics warehouse that runs SQL over columnar storage, separating compute from storage so queries scan only referenced columns and cost scales with the data read.

A serverless compute option for running single-purpose code in response to HTTP requests or events from Pub/Sub, Cloud Storage, and Eventarc, with the platform handling scaling and availability.