← Google Associate Cloud Engineer
Test yourself →

Setting up a cloud environment

Setting up a cloud environment

Before you touch any GCP resource you need the right structure underneath it: an Organisation, Folders, Projects, and Billing Accounts, all tied together with Identity and Access Management (IAM).

The resource hierarchy

  • Organisation sits at the top, tied to a Google Workspace or Cloud Identity domain.
  • Folders group Projects (eg by department or environment) and can be nested inside other Folders.
  • Projects are the base-level container for all resources - every resource belongs to exactly one Project.
  • Every Project has three identifiers: a Project ID (globally unique, immutable, user-chosen or auto-generated), a Project Name (mutable, not unique), and a Project Number (auto-assigned, immutable, globally unique).
  • IAM policies set higher in the hierarchy (Org or Folder) are inherited downward and combined with policies set lower down - permissions only ever add up, they never get revoked by a lower-level policy.

Billing

  • A Billing Account must be linked to a Project before you can create billable resources.
  • Billing Accounts are either self-serve (credit card, individual) or invoiced (for larger organisations).
  • One Billing Account can fund multiple Projects; use Budgets and Alerts to get notified at set spend thresholds (they warn - they do not automatically stop billing).
  • Billing export to BigQuery is the standard way to analyse detailed cost data.

Setting up access - the gcloud CLI and Cloud SDK

  • Install the Cloud SDK, then run 'gcloud init' to authenticate and set your default Project, region and zone.
  • 'gcloud auth login' authenticates a human user; 'gcloud auth application-default login' sets credentials for local apps/SDKs to use (Application Default Credentials).
  • Service accounts authenticate via a downloaded JSON key or, better, attached identity (no key file) - avoid exporting keys where possible, it is a common exam trap and a real security risk.

Regions, zones, and quotas

  • A region is a specific geographical location (eg europe-west2); a zone is an isolated deployment area within a region (eg europe-west2-a). Zones in the same region have low-latency, high-bandwidth links.
  • Resources are zonal, regional, or global - Compute Engine VMs and disks are zonal; some load balancers are global.
  • Every Project starts with default quotas (eg CPUs per region) - these are soft limits you can request increases for via the console, they are not hard architectural ceilings.

Common mistakes

  • Forgetting that a new Project has no Billing Account linked, so API calls fail with a billing error.
  • Confusing Project ID with Project Number in scripts and Terraform.
  • Assuming IAM roles granted at Project level also need re-granting at Folder level - they are already inherited.
  • Leaving default 'Editor' role assigned to service accounts instead of scoping to least-privilege predefined or custom roles.
  • Every GCP resource belongs to exactly one Project, and every Project belongs to at most one Organisation.
  • A Project has three identifiers: Project ID (immutable, globally unique), Project Name (mutable), and Project Number (auto-assigned, immutable).
  • Folders can be nested inside other Folders and group Projects for shared IAM policy and billing management.
  • IAM policies inherit downward through Org to Folder to Project to resource, and permissions are additive - never revoked by a lower level.
  • A Billing Account must be linked to a Project before billable resources can be created.
  • One Billing Account can be linked to many Projects; Budgets and Alerts notify on spend but do not auto-stop billing.
  • 'gcloud init' sets up authentication plus default Project, region and zone for the CLI.
  • 'gcloud auth application-default login' sets Application Default Credentials for local apps and SDKs, separate from user login.
  • A region (eg europe-west2) contains multiple zones (eg europe-west2-a, -b, -c) with low-latency links between them.
  • Resources are scoped as zonal, regional, or global depending on the service (eg VMs are zonal, some load balancers are global).
  • New Projects start with default resource quotas (eg CPUs per region) that are soft limits and can be raised on request.
  • Service accounts should avoid exported JSON keys where possible - attached identity is the more secure pattern the exam favours.
What are the four levels of the GCP resource hierarchy, top to bottom?
Organisation, Folder, Project, Resource.
tap to reveal
Which Project identifier is globally unique and immutable but chosen by the user at creation?
Project ID.
tap to reveal
Which Project identifier is auto-assigned by Google and never changes?
Project Number.
tap to reveal
Do IAM permissions set at a higher level in the hierarchy get overridden by a more restrictive policy lower down?
No - IAM is additive; permissions inherited from above are combined with, not revoked by, lower-level policies.
tap to reveal
What must be linked to a Project before you can create billable resources?
A Billing Account.
tap to reveal
Can one Billing Account fund multiple Projects?
Yes.
tap to reveal
Do Budget Alerts automatically stop billing when a threshold is hit?
No - they only send a notification; you must act on it or set up separate automation to stop spend.
tap to reveal
What command authenticates the gcloud CLI and sets default Project, region and zone?
gcloud init.
tap to reveal
What is the difference between 'gcloud auth login' and 'gcloud auth application-default login'?
'auth login' authenticates you as a human user for gcloud commands; 'application-default login' sets Application Default Credentials for local apps and SDKs to use.
tap to reveal
What is the relationship between a region and a zone?
A region is a geographic area (eg europe-west2) made up of multiple isolated zones (eg europe-west2-a) with low-latency, high-bandwidth connections between them.
tap to reveal
Are Compute Engine VMs zonal, regional, or global resources?
Zonal.
tap to reveal
Are default GCP quotas (eg CPUs per region) hard architectural limits?
No - they are soft limits you can request an increase for via the console.
tap to reveal
What is the more secure alternative to downloading a service account JSON key?
Attaching the service account identity directly to the resource (no exported key file).
tap to reveal
Can Folders be nested inside other Folders?
Yes.
tap to reveal
Where does Google recommend exporting detailed billing data for analysis?
BigQuery, via billing export.
tap to reveal

Planning & configuring a solution

Planning and configuring a GCP solution

Before you touch the console or gcloud, the ACE exam wants you to think about project structure, resource hierarchy, and billing. Everything in Google Cloud sits inside a hierarchy: Organization > Folders > Projects > Resources. Folders group projects (e.g. by department or environment) and IAM policies inherit downward. A resource always belongs to exactly one project, and every project has a unique Project ID (immutable, globally unique) plus a Project Number (auto-assigned) and a Project Name (editable, not unique).

Choosing compute

Part of planning is picking the right compute product: Compute Engine (VMs, full control), Google Kubernetes Engine (containers, orchestration), App Engine (fully managed PaaS, zero infra), Cloud Run (managed containers, scales to zero), and Cloud Functions (event-driven, single purpose). The exam loves scenario questions - match workload needs (stateful vs stateless, scaling to zero, custom runtime) to the right service.

Billing accounts and budgets

A billing account must be linked to a project before you can use most services. Billing accounts can be self-serve (credit card) or invoiced. Budgets and alerts are set per billing account or per project, with configurable threshold percentages (e.g. 50%, 90%, 100%) - alerts notify but do NOT automatically stop spending unless you build automation (e.g. a Cloud Function triggered by a Pub/Sub budget notification that disables billing).

Quotas

GCP enforces two types of quotas: rate quotas (reset after a time interval, e.g. API calls per 100 seconds) and allocation quotas (fixed resource counts, e.g. number of VPCs per project). Quotas are checked at project and region level. You can request quota increases via the console (IAM & Admin > Quotas) - some increases are automatic, others need Google approval and can take days.

Regions, zones, and resource scope

Resources are either global (e.g. images, snapshots, VPC networks), regional (e.g. subnets, static external IPs), or zonal (e.g. VM instances, persistent disks, GKE zonal clusters). A key planning mistake is placing a VM in one zone but its persistent disk in another - they must match. Choose regions based on latency to users, price, and compliance/data-residency needs.

Common exam traps

  • Confusing Project ID (immutable) with Project Name (editable).
  • Forgetting that deleting a project doesn't free the Project ID for reuse.
  • Assuming budget alerts stop spending automatically - they don't.
  • Mixing zonal and regional resources incorrectly.
  • Not knowing that Organization Policies constrain what resources CAN be created (allow/deny lists) regardless of IAM permissions.
  • Resource hierarchy is Organization > Folders > Projects > Resources, with IAM policies inheriting downward.
  • A Project ID is globally unique and immutable for the life of the project - it can never be reused after deletion.
  • A billing account must be linked to a project before most paid services can be used.
  • Budget alerts notify at chosen thresholds but do NOT automatically stop spending without custom automation.
  • GCP quotas come in two flavours: rate quotas (reset over time, e.g. per 100 seconds) and allocation quotas (fixed counts).
  • Resources are scoped as global (images, VPC networks), regional (subnets, static IPs), or zonal (VM instances, persistent disks).
  • A VM instance and its attached persistent disk must be in the same zone.
  • Organization Policies constrain what CAN be created (allow/deny constraints), separate from and on top of IAM permissions.
  • Cloud Run scales to zero and is billed only while handling requests, unlike always-on Compute Engine VMs.
  • App Engine is fully managed PaaS with zero infrastructure management, ideal for standard web app scaling scenarios.
  • Folders can group projects by department, team, or environment (e.g. prod vs dev) and pass down IAM policy.
What is the correct order of the GCP resource hierarchy?
Organization > Folders > Projects > Resources
tap to reveal
Is a Project ID mutable or immutable?
Immutable - it is set at creation and can never be changed or reused, even after project deletion
tap to reveal
Do budget alerts automatically stop billing when a threshold is hit?
No - they only notify; automatic action (like disabling billing) requires custom automation via Pub/Sub and Cloud Functions
tap to reveal
What are the two types of GCP quotas?
Rate quotas (reset after a time interval) and allocation quotas (fixed resource counts)
tap to reveal
Name three examples of global resources in GCP.
Images, snapshots, and VPC networks
tap to reveal
Name two examples of zonal resources in GCP.
VM instances and persistent disks
tap to reveal
Must a VM and its persistent disk be in the same zone?
Yes, always
tap to reveal
Which compute product scales to zero and only bills while handling requests?
Cloud Run
tap to reveal
Which compute product is fully managed PaaS requiring zero infrastructure management?
App Engine
tap to reveal
What must be linked to a project before most paid GCP services can be used?
A billing account
tap to reveal
What do Organization Policies control, as distinct from IAM?
What resources CAN be created (allow/deny constraints), regardless of a user's IAM permissions
tap to reveal
What are the three parts of a GCP project's identity?
Project ID (unique, immutable), Project Number (auto-assigned), and Project Name (editable, not unique)
tap to reveal
What can Folders be used for in planning a solution?
Grouping projects by department, team, or environment, with IAM policy inheritance flowing down to contained projects
tap to reveal
Give an example of a regional (not zonal, not global) resource.
A subnet or a static external IP address
tap to reveal

Deploying & implementing (Compute, GKE, App Engine)

Compute Engine deployment basics

Compute Engine gives you full control over VMs. You choose machine type, boot disk image, network, and startup scripts. Use instance templates to define a reusable VM config, then instance groups to deploy at scale.

  • Managed Instance Groups (MIGs) support autoscaling, autohealing (via health checks), rolling updates, and load balancing across zones.
  • Unmanaged instance groups let you group dissimilar VMs but cannot autoscale or autoheal — rarely the right exam answer.
  • Preemptible VMs run up to 24 hours max and can be reclaimed anytime with 30 seconds notice; Spot VMs are the newer version with no fixed 24-hour cap. Both are much cheaper but unsuitable for stateful, uninterruptible workloads.
  • Startup scripts (via metadata key startup-script) run on every boot — handy for bootstrapping without baking custom images.

GKE (Google Kubernetes Engine)

GKE runs containerised workloads on managed Kubernetes clusters.

  • Two cluster modes: Standard (you manage node config, full control) and Autopilot (Google manages nodes, you pay per pod resource — recommended default for new clusters since it's hands-off and cost-efficient).
  • Node pools group nodes with the same config; you can have multiple pools in one cluster for different workload needs (eg GPU pool vs standard pool).
  • Cluster autoscaler adjusts node count; Horizontal Pod Autoscaler (HPA) adjusts pod replica count based on CPU/metrics — know the difference, they're commonly confused.
  • Deployments manage stateless pods with rolling updates; StatefulSets are for stateful workloads needing stable identity/storage.
  • Use gcloud container clusters create for cluster creation and kubectl for workload management once you have credentials via gcloud container clusters get-credentials.

App Engine

App Engine is a fully managed, serverless platform for deploying application code without managing servers.

  • Standard environment: sandboxed, fast autoscaling (including scale-to-zero), supports specific language runtimes, cheaper for spiky/light traffic.
  • Flexible environment: runs in Docker containers on Compute Engine VMs, supports any language/custom runtime, no scale-to-zero, slower scaling, higher baseline cost.
  • Deploy with gcloud app deploy, using app.yaml to configure the service, scaling type, and runtime.
  • Traffic splitting lets you route percentages of traffic between versions for gradual rollouts or A/B testing — a favourite exam scenario.

Common exam traps

  • Don't pick Compute Engine when the question hints at 'no infrastructure management' — that's App Engine or Cloud Run territory.
  • Don't confuse regional vs zonal MIGs: regional MIGs spread instances across zones for higher availability.
  • Remember preemptible/Spot VMs are unsuitable for anything needing guaranteed uptime or long single-session jobs.
  • Autopilot GKE removes node management entirely — pick it whenever the scenario says 'minimise operational overhead' for Kubernetes.
  • Preemptible VMs have a hard 24-hour maximum runtime and can be reclaimed with only 30 seconds warning.
  • Spot VMs are the modern replacement for preemptible VMs and have no fixed 24-hour runtime cap.
  • Managed Instance Groups (MIGs) support autoscaling, autohealing, and rolling updates; unmanaged groups support none of these.
  • Regional MIGs distribute instances across multiple zones in a region for higher availability than zonal MIGs.
  • GKE Autopilot mode has Google manage node infrastructure entirely; you pay per pod resource requested.
  • Horizontal Pod Autoscaler (HPA) scales pod replica count; Cluster Autoscaler scales the number of nodes in the cluster.
  • App Engine Standard supports scale-to-zero and fast autoscaling but only specific language runtimes in a sandbox.
  • App Engine Flexible runs in Docker containers on Compute Engine VMs, supports any runtime, but never scales to zero.
  • App Engine deployments are configured via app.yaml and pushed with the gcloud app deploy command.
  • Traffic splitting in App Engine lets you route a percentage of live traffic to a new version for gradual rollout.
  • Startup scripts on Compute Engine are set via the startup-script metadata key and run on every VM boot.
  • kubectl talks to GKE only after running gcloud container clusters get-credentials to fetch cluster credentials.
What is the maximum runtime for a preemptible VM?
24 hours, after which it is automatically terminated.
tap to reveal
How much warning does a preemptible or Spot VM get before reclamation?
30 seconds.
tap to reveal
What is the key difference between preemptible and Spot VMs?
Spot VMs have no fixed 24-hour runtime limit; preemptible VMs do.
tap to reveal
What can a Managed Instance Group do that an unmanaged one cannot?
Autoscale, autoheal via health checks, and perform rolling updates.
tap to reveal
What is a regional MIG and why use one?
A MIG that spreads instances across multiple zones in a region, for higher availability than a zonal MIG.
tap to reveal
What does GKE Autopilot mode manage for you compared to Standard mode?
It fully manages node infrastructure; you only manage workloads and pay per pod resource.
tap to reveal
What is the difference between HPA and Cluster Autoscaler in GKE?
HPA scales the number of pod replicas; Cluster Autoscaler scales the number of nodes in the cluster.
tap to reveal
When would you use a StatefulSet instead of a Deployment in GKE?
When pods need stable network identity and persistent storage, ie for stateful workloads.
tap to reveal
What are the two App Engine environments and their key trade-off?
Standard (sandboxed, fast scale-to-zero, limited runtimes) and Flexible (Docker on Compute Engine VMs, any runtime, no scale-to-zero).
tap to reveal
What file configures an App Engine service's runtime and scaling?
app.yaml.
tap to reveal
What command deploys an App Engine application?
gcloud app deploy
tap to reveal
What GKE feature lets you route a percentage of traffic to a new App Engine version?
Traffic splitting.
tap to reveal
What metadata key runs a script on every Compute Engine VM boot?
startup-script
tap to reveal
What command fetches credentials so kubectl can manage a GKE cluster?
gcloud container clusters get-credentials
tap to reveal
Why would 'minimise operational overhead' in a Kubernetes exam scenario point to a specific GKE mode?
It points to GKE Autopilot, since Google manages node infrastructure entirely.
tap to reveal

Configuring access & security (IAM)

IAM: the basics

Google Cloud IAM controls who (identity) can do what (role) on which resource. Every permission check follows this identity + role + resource pattern. You almost never grant permissions directly - you grant roles, and roles bundle permissions together.

Identities can be Google accounts, service accounts, Google groups, Cloud Identity domains, or 'allUsers'/'allAuthenticatedUsers' for public access. Prefer groups over individual users for easier management at scale.

Role types

  • Basic roles: Owner, Editor, Viewer. Broad, legacy, apply project-wide - avoid in production, too permissive.
  • Predefined roles: curated by Google per service, e.g. roles/compute.instanceAdmin. Best practice for most cases.
  • Custom roles: you define the exact permission set. Useful for least privilege but need maintenance as APIs change.

The resource hierarchy

Policies flow downward: Organisation > Folder > Project > Resource. A role granted at the organisation level is inherited by every folder, project and resource beneath it. You cannot restrict a permission lower down that was granted higher up - only add more. Deny policies (a separate feature) can block access regardless of allow policies, and deny always wins over allow.

Service accounts

A service account is an identity for applications/VMs, not a person. Each has an email like PROJECT_NUMBER-compute@developer.gserviceaccount.com (default Compute Engine SA) or name@project.iam.gserviceaccount.com. Avoid using the default service account with the broad Editor role in production - assign a custom SA with minimal roles instead. Service account keys (JSON) are a security risk if leaked - prefer attaching SAs directly to resources or using Workload Identity Federation instead of downloading keys.

Common mistakes to know for the exam

  • Confusing Owner (full control incl. billing/IAM) with Editor (modify resources, no IAM/billing changes) with Viewer (read-only).
  • Forgetting that IAM policies are additive across the hierarchy - there is no explicit lower-level 'deny' in a normal allow policy.
  • Granting roles at project level when only one resource needs access - always scope to the smallest level that works.
  • Leaving default service accounts with Editor role attached to VMs.
  • Not knowing that Cloud IAM Recommender can suggest removing unused/excess permissions.

Useful tools

  • Policy Analyzer / Policy Troubleshooter: work out why a user has or lacks access.
  • IAM Conditions: add attribute-based conditional access (e.g. time-bound or resource-tag-bound grants) to a role binding.
  • Organisation Policy Service: restricts what CAN be configured (e.g. disable external IPs) - separate from IAM, which controls WHO can act.
  • IAM policy = identity + role + resource; roles bundle permissions, you rarely grant permissions individually
  • Three basic roles: Owner (full control + billing/IAM), Editor (modify resources), Viewer (read-only) - avoid basic roles in production
  • Predefined roles are Google-curated per service (e.g. roles/compute.instanceAdmin.v1) and are the recommended default choice
  • Custom roles let you define an exact permission set for least privilege but require ongoing maintenance
  • Resource hierarchy is Organisation > Folder > Project > Resource; IAM policies are inherited downward and are additive only
  • Deny policies override allow policies wherever they apply, regardless of hierarchy level
  • Default Compute Engine service account email format is PROJECT_NUMBER-compute@developer.gserviceaccount.com
  • Avoid downloading service account JSON keys where possible - prefer attaching the SA to the resource or Workload Identity Federation
  • allUsers grants public/anonymous access; allAuthenticatedUsers grants access to any Google-authenticated account - both are high risk
  • Always grant the least privilege role at the smallest applicable scope in the resource hierarchy
  • IAM Recommender analyses actual usage and suggests removing excess or unused role grants
  • IAM Conditions allow attribute-based, conditional role bindings, e.g. time-bound or restricted-by-resource-tag access
What three basic roles exist in Cloud IAM?
Owner, Editor, Viewer
tap to reveal
Which basic role includes billing and IAM management rights?
Owner
tap to reveal
What is the resource hierarchy order in Google Cloud, top to bottom?
Organisation > Folder > Project > Resource
tap to reveal
Do IAM policies granted higher in the hierarchy get inherited lower down?
Yes, they are inherited and cannot be restricted lower down, only added to
tap to reveal
What wins when an allow policy and a deny policy conflict?
The deny policy always wins
tap to reveal
What is a service account used for?
It is an identity for applications, VMs or workloads to authenticate and act, not for a human user
tap to reveal
What is the default Compute Engine service account email format?
PROJECT_NUMBER-compute@developer.gserviceaccount.com
tap to reveal
Why should you avoid downloading service account JSON keys?
Keys can leak and grant persistent access; prefer attaching the SA to the resource or using Workload Identity Federation
tap to reveal
What does the identity allUsers grant?
Public, anonymous access to anyone on the internet
tap to reveal
What does allAuthenticatedUsers grant?
Access to anyone with any Google account, not just your organisation
tap to reveal
When should you use a predefined role over a custom role?
By default - predefined roles are Google-curated and cover most needs; use custom roles only when you need tighter least-privilege control
tap to reveal
What tool suggests removing excess or unused IAM permissions?
IAM Recommender
tap to reveal
What feature lets you add time-bound or attribute-based restrictions to a role binding?
IAM Conditions
tap to reveal
What is the best-practice scope level to grant an IAM role?
The smallest scope in the hierarchy that satisfies the requirement (least privilege)
tap to reveal
What separate service restricts what configurations are allowed, rather than controlling who can act?
Organisation Policy Service
tap to reveal

Networking & storage

VPC basics

A VPC network is global, but subnets are regional and span all zones in that region.

VPC networks come in two flavours: auto mode (one subnet per region, created automatically, /20 ranges) and custom mode (you define subnets and ranges yourself). Custom mode is best practice for production.

Subnets have a primary IP range plus optional secondary ranges, used for GKE alias IPs (pods and services).

Firewall rules are stateful, apply at the VPC level (not per subnet), and use priority numbers from 0-65535 (lower number = higher priority). Default rules: implied allow egress, implied deny ingress.

Connecting networks

  • VPC Peering: private RFC 1918 connectivity between two VPCs, no transitive peering, works across projects/orgs.
  • Shared VPC: a host project shares subnets with service projects, letting centrally managed networking span teams - common exam scenario for centralised IAM/network control.
  • Cloud VPN: encrypted tunnel over the public internet, quick to set up, up to 3 Gbps per tunnel.
  • Cloud Interconnect: Dedicated (10 or 100 Gbps direct connection) or Partner (50 Mbps-50 Gbps via a partner) for private, high-throughput, low-latency links to on-prem.
  • Cloud NAT lets private instances reach the internet outbound without external IPs; it does not allow inbound connections.

Load balancing

Know the split: global external load balancers (HTTP(S), SSL Proxy, TCP Proxy) front global backends and need Premium Tier networking; regional load balancers (Internal HTTP(S), Internal TCP/UDP, external regional Network LB) stay within one region.

HTTP(S) Load Balancing is Layer 7 and can route by URL path/host; Network Load Balancing is Layer 4 (pass-through).

Storage services and classes

Cloud Storage buckets are globally unique by name and hold objects, not filesystems. Four storage classes by access pattern, not performance: Standard (frequent access), Nearline (accessed <1x/month, 30-day minimum), Coldline (<1x/quarter, 90-day minimum), Archive (<1x/year, 365-day minimum). Early deletion before the minimum duration still incurs the full minimum-duration charge.

Use Object Lifecycle Management rules to auto-transition or delete objects by age/class/version.

Bucket location can be region, dual-region, or multi-region - multi-region gives higher availability, region gives lowest latency to a single area.

Persistent Disk vs Filestore vs Cloud SQL

Persistent Disk (zonal or regional) attaches to Compute Engine VMs like a virtual hard drive - zonal PD survives one zone only, regional PD replicates across two zones for HA.

Filestore provides managed NFS file storage for multiple VMs needing shared file access.

Cloud SQL is managed relational (MySQL/PostgreSQL/SQL Server); Cloud Spanner is for globally distributed, horizontally scalable relational needs; Firestore/Bigtable are NoSQL (Firestore = document/mobile-web, Bigtable = huge analytical/time-series workloads).

Common exam traps

  • Confusing IAM 'roles/storage.objectViewer' (bucket-level read) with ACLs (object-level, legacy, avoid unless required).
  • Forgetting Cloud NAT needs a Cloud Router in the same region.
  • Picking Nearline for daily-access data (wrong - use Standard) or Archive for monthly backups (too aggressive - use Coldline).
  • Assuming firewall rules are per-subnet; they are actually per-VPC network with target tags/service accounts scoping them.
  • VPC networks are global; subnets are regional and span all zones in that region
  • Auto mode VPCs create one /20 subnet per region automatically; custom mode requires manual subnet creation
  • Firewall rules are stateful and evaluated by priority 0-65535, lower number wins first
  • Default VPC firewall behaviour: implied allow all egress, implied deny all ingress
  • Cloud NAT provides outbound-only internet access for private instances and requires a Cloud Router
  • Dedicated Interconnect offers 10 Gbps or 100 Gbps direct private links to Google's network
  • Cloud Storage has 4 classes: Standard, Nearline (30-day min), Coldline (90-day min), Archive (365-day min)
  • Deleting an object before its storage class minimum duration still bills for the full minimum period
  • Regional Persistent Disk replicates synchronously across two zones for high availability; zonal PD does not
  • Filestore is Google's managed NFS solution for shared file storage across multiple Compute Engine VMs
  • HTTP(S) Load Balancing operates at Layer 7 and can route traffic based on URL path or host header
  • VPC Peering provides private connectivity between VPCs but is never transitive across a peering chain
Are Cloud Storage buckets regional or global in namespace?
Bucket names are globally unique across all of Google Cloud, though the bucket's data location can be region, dual-region, or multi-region
tap to reveal
What is the minimum storage duration for Nearline storage?
30 days - early deletion still bills for the full 30 days
tap to reveal
What is the minimum storage duration for Coldline storage?
90 days
tap to reveal
What is the minimum storage duration for Archive storage?
365 days
tap to reveal
What does Cloud NAT allow and not allow?
Allows private instances outbound internet access; does not allow unsolicited inbound connections
tap to reveal
What extra resource does Cloud NAT require in the same region?
A Cloud Router
tap to reveal
Difference between zonal and regional Persistent Disk?
Zonal PD lives in one zone only; regional PD synchronously replicates across two zones for HA
tap to reveal
What is Filestore used for?
Managed NFS file storage shared across multiple Compute Engine VMs
tap to reveal
Is VPC Peering transitive?
No - if VPC A peers with B and B peers with C, A cannot reach C through B
tap to reveal
What is Shared VPC used for?
Letting a host project centrally share subnets/networking with multiple service projects across teams
tap to reveal
OSI layer of HTTP(S) Load Balancing vs Network Load Balancing?
HTTP(S) LB is Layer 7 (application, can route by URL/host); Network LB is Layer 4 (pass-through)
tap to reveal
Bandwidth range of Dedicated Interconnect?
10 Gbps or 100 Gbps per connection, direct to Google's network
tap to reveal
Bandwidth range of Partner Interconnect?
50 Mbps up to 50 Gbps, via a supported service partner
tap to reveal
Default priority ordering rule for VPC firewall rules?
Lower priority number = evaluated first / higher precedence, range 0-65535
tap to reveal
Which storage class suits data accessed a few times a year, like yearly compliance backups?
Archive storage class
tap to reveal