Google ACE Glossary

Key terms and acronyms for the ACE exam, explained in plain language.

By Abu Anns · Last updated 20 November 2026Exam version: ACE (exam guide live 30 June 2026)

Access scopes
A legacy mechanism that limits what APIs a VM's service account can call. Scopes can only restrict permissions; IAM roles grant them. Modern best practice is to set broad scopes and control access with IAM roles.
AlloyDB
Google's PostgreSQL-compatible managed database for demanding workloads, offering higher performance and availability than Cloud SQL for PostgreSQL.
Artifact Registry
Google Cloud's registry for storing container images and other build artifacts. The successor to the deprecated Container Registry.
Autopilot (GKE)
A GKE mode in which Google manages the cluster nodes and bills for the Pod resources requested, rather than for the nodes themselves.
Backend service
A load balancing component that groups backends (instance groups, network endpoint groups, or serverless backends) and holds configuration such as the health check and session affinity.
BigQuery
Google Cloud's serverless data warehouse for large-scale SQL analytics. Charged by data stored and data processed by queries; not a transactional database.
Bigtable
A wide-column NoSQL database for massive-scale, high-throughput workloads such as time-series data and IoT telemetry.
Billing account
The payment instrument for Google Cloud usage. One billing account can pay for many projects; each project links to exactly one billing account.
Budget
A spending threshold with alert rules. Budgets only send alerts — they never stop spending by themselves.
Cloud CDN
Google's content delivery network, which caches content at edge locations worldwide to reduce latency. Enabled on an Application Load Balancer.
Cloud Functions
Google Cloud's serverless functions service: short, stateless units of code that run in response to events such as Pub/Sub messages or Cloud Storage object changes.
Cloud Identity
Google's identity service for organizations without Google Workspace, providing managed users and groups under the company domain.
Cloud NAT
A service that lets VMs without external IP addresses initiate outbound internet connections while remaining unreachable from the internet.
Cloud Run
A fully managed serverless platform for running containers. Scales automatically to zero and bills only for request-handling time.
Cloud SQL
A fully managed relational database service for MySQL, PostgreSQL, and SQL Server.
Cloud VPN
An IPsec VPN service connecting a VPC to an on-premises network or another cloud over the internet. HA VPN provides two tunnels with automatic failover.
Custom machine type
A Compute Engine VM shape with an exact vCPU and memory combination chosen by the user, instead of a predefined machine type.
Custom role
An IAM role defined by the user with exactly the permissions a job needs, used when no predefined role fits.
Dataflow
Google Cloud's managed batch and streaming data-processing service, based on Apache Beam.
Eventarc
Google Cloud's event router, delivering events from Google Cloud services and custom sources to targets such as Cloud Run, with filtering.
Firewall rule
A VPC-level rule controlling traffic to and from VM instances. Stateful, with a direction (ingress/egress), priority, action, and target.
Firestore
A NoSQL document database for mobile and web applications, with real-time synchronization to clients.
GKE (Google Kubernetes Engine)
Google's managed Kubernetes service for running containerized applications, in Autopilot or Standard mode.
HA VPN
High-availability Cloud VPN: two IPsec tunnels with automatic failover and a 99.99% availability SLA.
Health check
A probe that determines whether a load balancer backend is healthy enough to receive traffic. Unhealthy backends are removed from rotation automatically.
HPA (Horizontal Pod Autoscaler)
A Kubernetes mechanism that changes the number of Pod replicas based on CPU, memory, or a custom metric.
IAM policy
A collection of bindings that grant roles to principals on a resource. Policies attach to the organization, folder, project, or resource level and are inherited downward.
Instance template
An immutable definition of a VM (machine type, image, disks, network) used to create VMs in a managed instance group.
Log-based metric
A Cloud Monitoring metric derived from log entries matching a filter — for example, counting ERROR entries per minute. Used to alert on log patterns.
Log sink
A Cloud Logging routing rule that sends matching log entries to a destination: Cloud Storage, BigQuery, Pub/Sub, or another project.
Managed instance group (MIG)
A group of identical VMs created from an instance template, with autoscaling, autohealing, and rolling updates. Regional MIGs span zones.
Network Service Tiers
A choice of network quality for egress traffic: Premium Tier routes on Google's private backbone; Standard Tier hands traffic to the public internet early and is cheaper.
Ops Agent
The unified agent installed on VMs to collect metrics and logs for Cloud Monitoring and Cloud Logging.
Organization policy
A guardrail set through the Organization Policy Service that restricts what resources are allowed to do, inherited down the resource hierarchy. Cannot be overridden by IAM.
OS Login
A Compute Engine feature that ties SSH access to IAM, so users log in with their Google identity and roles control who can log in and sudo.
Persistent Disk
Block storage for Compute Engine VMs and GKE nodes. Zonal disks live in one zone; regional disks replicate synchronously across two zones.
Predefined role
A Google-maintained IAM role for a specific job function, such as Compute Viewer or Storage Object Creator. The default choice for least-privilege access.
Pub/Sub
Google Cloud's asynchronous messaging service. Publishers send messages to topics; subscribers receive them via pull, push, or export subscriptions.
Service account
An identity for a machine or application, used to authenticate API calls from workloads running on Google Cloud or elsewhere.
Service account impersonation
Temporarily acting as a service account using short-lived credentials, without holding its key.
Shared VPC
A VPC network owned by a host project and shared with service projects, so one team manages the network while others deploy into it.
Snapshot
A point-in-time backup of a Persistent Disk. Incremental, schedulable, and usable to create new disks or images.
Spanner
Google's globally distributed relational database, providing strong consistency and horizontal scaling across regions.
Spot VM
Spare Compute Engine capacity sold at a 60–91% discount, reclaimable by Google at any time with 30 seconds of warning. For fault-tolerant, interruptible workloads.
VPC Network Peering
A private connection between two VPC networks so resources can communicate using internal IP addresses. Non-transitive.
VPA (Vertical Pod Autoscaler)
A Kubernetes mechanism that adjusts the CPU and memory requests and limits of Pods based on observed usage.
Workload Identity Federation
A way for workloads outside Google Cloud (AWS, on-premises, GitHub Actions) to authenticate with short-lived tokens using their own identity provider, without service account keys.
Advertisement