Prepare and test your skills
Prepare and test your skills
Hybrid networking connects an on-premises network to a Google Cloud Virtual Private Cloud (VPC) network so traffic can move securely between the two environments. The connection uses either encrypted tunnels over the internet or private circuits, with dynamic routing to exchange routes automatically. Production designs add redundancy to meet availability agreements.
Cloud VPN (opens in a new tab) is the Network Connectivity product that creates encrypted tunnels over the public internet between a VPC network and an on-premises network or another cloud. Traffic flows from the on-premises network through the encrypted tunnel to the VPC network, and return traffic flows back through the same tunnel. High Availability (HA) VPN is the production-ready form with a 99.99% service level agreement (SLA), which shifts uptime responsibility from a customer-managed virtual machine (VM) VPN endpoint to Google infrastructure.
An intermediary VM as a VPN endpoint creates a single point of failure and operational burden. The customer must maintain, secure, and keep that VM available, and one VM failure can block all communication between environments. A managed connectivity product removes that VM from the application path so availability is handled by Google.
Cloud Router manages Border Gateway Protocol (BGP) routing between the VPC network and the on-premises router over Cloud VPN or Cloud Interconnect. Cloud Router advertises VPC routes outward and learns on-premises routes inward, so new subnets propagate without manual static routes. The tunnels provide encrypted transport while Cloud Router provides the control plane, and if the BGP session is not in the UP state, routes are not exchanged and traffic cannot be forwarded.
Redundant hybrid topologies use duplicated managed paths so no single circuit failure breaks the routed connection between on-premises and the VPC network. The production-level 99.99% monthly uptime configuration requires four Cloud Interconnect connections across two Google Cloud regions. Connection pairs must sit in different metropolitan areas and in separate edge availability domains, with Cloud Router advertising routes to the on-premises router.
On-premises routers in two metros connect over four Interconnect links to Cloud Routers in two Google Cloud regions, with BGP exchanging routes into a shared VPC and failover across remaining paths.
Traffic flows from the on-premises router over the Interconnect connections to Cloud Router, and Cloud Router advertises VPC routes back. If one metro, edge domain, or region is impaired, traffic continues over the remaining paths to the other region. Google recommends using interconnect connection groups to verify that the design meets the intended availability level.
| Topology option | Composition | Placement rule |
|---|---|---|
| Four Dedicated connections | Four Dedicated Interconnect connections | Across two regions, different metros and separate edge availability domains |
| Four Partner attachments | Four Partner Interconnect VLAN attachments | Across two regions, different metros and separate edge availability domains |
| Mixed Dedicated and Partner | Two Dedicated connections in one region plus two Partner attachments in another region | Across two regions, different metros and separate edge availability domains |
Choose four Dedicated connections when direct colocation access is available, four Partner attachments when provider connectivity is needed, and the mixed option when the two regions have different access constraints. Single-circuit designs do not qualify because a single Dedicated connection and a single Partner VLAN attachment both carry No Uptime SLA and can be drained for hours or days during maintenance.
Exam tip: A single Dedicated Interconnect connection and a single Partner Interconnect VLAN attachment both carry No Uptime SLA; 99.99% requires four connections across two regions with different metros and separate edge availability domains.
Dedicated Interconnect and Partner Interconnect are the two private Cloud Interconnect options that extend on-premises to VPC networks without traversing the public internet. Dedicated Interconnect is a direct physical connection where traffic flows directly between networks, while Partner Interconnect flows through a supported service provider. Both provide access to Google Cloud products except Google Workspace and support Private Google Access to APIs from on-premises.
Dedicated Interconnect requires the organization to install routing equipment in a Google-supported colocation facility and configure BGP on both on-premises routers and Cloud Routers. Connection capacity uses 10-Gbps, 100-Gbps, or 400-Gbps Ethernet circuits with up to 8 circuits per connection, plus flexible VLAN attachments from 50 Mbps to 50 Gbps. Partner Interconnect needs no colocation equipment because the provider already connects to Google, and each VLAN attachment provides 50 Mbps to 50 Gbps with growth depending on provider capacity. For Partner Layer 2 connections the customer configures BGP on both sides, while for Layer 3 the Cloud Router configuration is automated. Deploy HA VPN over Interconnect to encrypt traffic for either option, with Google covering end-to-end SLA for Dedicated topologies and covering only the Google-to-provider segment for Partner.
Continue with this study guide: Extending to on-premises environments (hybrid networking)
Multicloud networking links Google Cloud VPC networks to on-premises networks, other clouds, and to each other with automatic route exchange and failover. Cloud Router with BGP advertises VPC subnets to peers and learns peer routes in return, so traffic can fail over between tunnels without static routes. Private methods then control how separate VPC networks share routes, subnets, or individual services.
Cloud Router (opens in a new tab) is a regional service that manages BGP sessions for Cloud VPN tunnels and VLAN attachments. Each Cloud Router belongs to one VPC network in one region and uses one Google autonomous system number (ASN) for all its sessions. The Google ASN must be a private ASN in the range 64512 through 65534 or 4200000000 through 4294967294 that is not already used in the same network, and the ASN cannot be changed after creation.
A BGP session runs over each VPN tunnel through a Cloud Router interface bound to that tunnel. Traffic flows from the VPC network to the Cloud Router interface, through the VPN tunnel to the peer gateway interface, and into the peer network. A Cloud Router used for VPN must not already manage a BGP session for a Partner Interconnect VLAN attachment and should not be shared with Cloud NAT. The peer must support BGP, and the VPC should use global routing mode so learned routes apply across regions.
An HA VPN gateway is a regional gateway with two interfaces, each with its own automatically assigned external IP address. A tunnel connects one HA VPN interface to one peer interface, with a BGP session on top of each tunnel. Building the connection follows an ordered dependency:
0 to peer interface 0 and a second from interface 1 to interface 1, using matching pre-shared keys and IKEv2.For Google Cloud-to-Google Cloud links, the same pattern is used on both sides with a Cloud Router in each VPC. Tunnel 0 must connect to interface 0 of the other gateway and tunnel 1 to interface 1, or the tunnels are not established. IPv6 requires IKEv2 and a dual-stack or IPv6 stack type that cannot be changed later.
External peer redundancy describes how many peer interfaces the external VPN gateway resource represents. The resource sits outside the Google trust boundary but is managed in the Google Cloud project, and the HA VPN gateway depends on it to map each tunnel to the correct peer IP.
| Peer configuration | Interfaces | Redundancy type |
|---|---|---|
| Single peer gateway with one IP | ID 0 | SINGLE_IP_INTERNALLY_REDUNDANT |
| Peer gateway with two IPs, or two gateways with one IP each | IDs 0 and 1 | TWO_IPS_REDUNDANCY |
| Peer gateways with four IPs, such as Amazon Web Services (AWS) | IDs 0, 1, 2, and 3 | FOUR_IPS_REDUNDANCY |
Advertised route priority controls whether both tunnels carry traffic at once. By default both BGP sessions advertise unmodified base priorities, creating an active-active configuration where traffic flows over both tunnels. An active-passive configuration uses unequal priorities, for example priority 10 on tunnel 1 and priority 20 on tunnel 2, so traffic prefers tunnel 1 and uses tunnel 2 only after failure. Custom advertisement can further limit prefixes by setting advertisement mode to custom with specific IP ranges.
Exam tip: A 99.99% agreement requires VPN tunnels on both HA VPN gateway interfaces; a single tunnel does not qualify even if BGP is configured.
Inter-VPC communication lets separate VPC networks exchange traffic while keeping isolation boundaries. A VPC network is an isolated private partition, so workloads in different VPC networks cannot communicate by default even though subnets in different regions of the same VPC communicate privately over Google's backbone. Enterprises separate VPCs to enforce default-deny between environments, to reuse overlapping IP ranges, and to delegate full network control to one team.
VPC Network Peering is a direct peering connection where workloads in two VPC networks communicate with internal IP addresses without traversing the internet. Peering merges route propagation so forwarding behaves as if VMs were in the same VPC, automatically exchanging subnet routes with optional exchange of reused public ranges and custom routes. Each peer keeps its own firewall and Identity and Access Management (IAM) policies, requires non-overlapping IPs, and is not transitive, so spokes peered to the same hub cannot reach each other through the hub alone.
Shared VPC lets one host project share a common VPC network with service projects in the same organization. Central administrators control subnets, routes, and firewalls in the host project, while service-project administrators deploy VMs into designated subnets. Private Service Connect is a consumer-to-producer model where a consumer VPC connects privately to a producer service in a separate VPC without merging networks, working within or across organizations.
| Method | Connectivity model | Works across organizations | Isolation boundary |
|---|---|---|---|
| VPC Network Peering | Two VPC networks exchange routes for VM-to-VM reachability | Yes, across projects and organizations | Each VPC keeps its own firewall and IAM; not transitive |
| Shared VPC | Service projects share subnets in one host-project VPC | No, inside one organization resource | Central network control with delegated workload deployment |
| Private Service Connect | Consumer VPC connects privately to a producer service | Yes, same or different organizations | Only the published service is exposed; networks not merged |
Hybrid and multicloud connectivity options differ by path, bandwidth, and whether the far end is on-premises or another cloud. All options pair with Cloud Router to exchange routes dynamically, applying learned external routes as custom dynamic routes in the VPC. Cross-Cloud Interconnect is the dedicated Google-provisioned link between a Google Cloud VPC and another provider, requiring a primary plus redundant pair per provider for AWS, Azure, Oracle Cloud Infrastructure, and Alibaba Cloud.
Cross-Cloud Interconnect delivers up to 100 Gbps per link with 99.9% and 99.99% high-availability options and supports Media Access Control Security (MACsec) for network-layer encryption. Cloud VPN extends a peer network through IPsec tunnels with lower cost but lower bandwidth and less deterministic performance than Interconnect. Dedicated Interconnect uses direct 10-Gbps, 100-Gbps, or 400-Gbps circuits up to 8 per connection with 50 Mbps to 50 Gbps VLAN attachments, while Partner Interconnect uses 50 Mbps to 50 Gbps VLAN attachments through a provider with no colocation equipment.
| Option | Network path | Capacity and setup |
|---|---|---|
| Cloud VPN | Private secure path over internet with IPsec between gateways | Lower bandwidth; choose for low cost or experimental use |
| Dedicated Interconnect | Direct connection to Google, no internet or provider | 10, 100, or 400-Gbps circuits up to 8 per connection; VLANs 50 Mbps to 50 Gbps; requires colocation equipment |
| Partner Interconnect | Dedicated path through supported provider, no internet | VLANs 50 Mbps to 50 Gbps; no colocation equipment |
| Cross-Cloud Interconnect | Dedicated physical link provisioned by Google to another cloud | Up to 100 Gbps per link; paired primary and redundant connections per provider |
Security protection in Google Cloud combines network firewalls, edge filtering, intrusion monitoring, and identity-based access. Traffic moves from the internet edge toward the VPC through successive controls, with each layer reducing load or inspecting what prior layers allowed. Private connectivity then keeps workloads off the public internet wherever possible.
VPC firewall rules (opens in a new tab) are project-level controls bound to a single VPC network, while hierarchical firewall policies attach at the organization or folder level and apply to many projects below. Traffic flows from sources outside the protected network to destinations inside it, and both layers are evaluated to create granular isolation. Central teams can block risky traffic everywhere while project teams open only needed workload paths.
A VPC rule targets workloads by IP ranges, network tags, or service accounts. IP ranges match by CIDR block, network tags match VMs carrying the tag, and service accounts match VMs running as a specific identity. Direction, priority, and matching decide when the rule applies, with lower priority numbers evaluated first. For example, an INGRESS rule with priority 1000 and source_ranges of 10.1.0.0/16 matches only traffic from the Private Service Connect NAT subnet, and an allow block with protocol = "all" permits all protocols and ports for that flow.
Firewall logging records connection metadata for auditing without changing the allow or deny decision. With log_config set to INCLUDE_ALL_METADATA, logs flow from the enforcement point to Cloud Logging for review of allowed or denied flows. For workload paths such as Site Recovery Manager (SRM), firewalls must permit SRM server to vCenter traffic and SRM server to peer SRM server traffic between protected and recovery sites.
Google Cloud Armor is an edge security service that filters traffic from the internet before it reaches workloads or enters VPC networks. Traffic flows from the internet to the Google edge and then to protected backends, with enforcement inside points of presence where external load balancing occurs. Cloud Intrusion Detection System (Cloud IDS) complements it as an additional monitoring layer in a multi-layer design.
Google Cloud Armor security policies are ordered collections of rules with a match condition and an action such as permit, deny, rate-limit, or redirect. Match conditions can use Layer 3 through Layer 7 attributes:
DDoS protection splits by layer because volumetric floods and application floods behave differently:
| Protection layer | Activation | What stops the attack |
|---|---|---|
| L3/L4 volumetric and protocol attacks | Always on for protected load balancers | Automatic inline detection and mitigation by Google |
| L7 application attacks and HTTP floods | Requires user-configured security policy | Custom filtering and Web Application Firewall (WAF) rules on backend services |
Baseline protection automatically covers global external, classic, and external proxy Network Load Balancers for HTTP, HTTPS, HTTP/2, and QUIC. Preconfigured WAF rules package OWASP ModSecurity Core Rule Set signatures for SQL injection, cross-site scripting, and command injection, supporting versions 4.22, 3.3, and 3.0 with 4.22 recommended. XML body parsing is not supported, and managed rules add automatically updated signatures for new CVEs. Enterprise adds threat intelligence feeds, Adaptive Protection for L7 anomalies, and advanced network DDoS monitoring.
Exam tip: L3/L4 DDoS mitigation is always on, while L7 HTTP-flood protection works only after you create a security policy and attach it to a backend service.
Layered defense combines edge blocking, reachability control, and inspection so no single control must stop every attack. Google Cloud Armor blocks unwanted traffic at the edge, Cloud Next Generation Firewall controls reachability, and Cloud IDS with intrusion prevention inspects deeper threats. Traffic therefore moves through policy enforcement first, then load balancing and firewall evaluation, then deeper inspection.
Attaching protection follows an ordered workflow where each step depends on the prior step:
Policies attach to external Application Load Balancers, regional internal Application Load Balancers, and global, classic proxy, and regional passthrough Network Load Balancers using schemes EXTERNAL, EXTERNAL_MANAGED, or INTERNAL_MANAGED. If multiple forwarding rules point to the same backend service, the same policy enforces traffic entering through each address.
Identity-based access grants access based on verified identity while keeping workloads off the public internet. Identity-Aware Proxy (IAP) checks Google identity and IAM permission before traffic reaches applications on App Engine, Compute Engine, and Google Kubernetes Engine (GKE). A user request flows to IAP first, and only allowed traffic flows from IAP to the internal application. For SSH, IAP TCP forwarding reaches a VM with no public IP without opening SSH to the world.
Private connectivity removes external addresses so traffic stays inside the VPC boundary. Supporting controls include restricting public IPs on Compute Engine, enabling private GKE clusters, enabling Private Google Access on subnets and instances, and enforcing public access prevention on Cloud SQL and Cloud Storage. Firewall policy still enforces the network boundary for IAP flows by allowing TCP port 22 from 35.235.240.0/20, which contains all IAP TCP-forwarding addresses. For hybrid links, choose Cloud VPN for lower cost and lower bandwidth and Cloud Interconnect for enterprise-grade higher throughput.
VPC design and load balancing determine how Google Cloud workloads reach each other, the internet, Google APIs, and on-premises or adjacent networks. A Virtual Private Cloud (VPC) (opens in a new tab) is a global private network with regional subnets, extended by sharing, peering, private endpoints, and hybrid links. Load balancers then distribute traffic across backends while private access designs control egress.
Cloud Load Balancing distributes traffic across backends such as instance groups, network endpoint groups (NEGs), and on-premises endpoints. Load balancer choices split by scope, reachability, and proxy layer. Scope is global versus regional, reachability is external internet-facing versus internal private, and proxy behavior is Layer 4 TCP/SSL proxying versus Layer 7 application routing with a URL map.
A global classic proxy Network Load Balancer with Premium Tier advertises a global anycast IP from all points of presence worldwide, with one backend service able to use backends in multiple regions. A regional balancer uses a regional external or internal IP advertised from points of presence for its region, and all components must stay in that region. The regional internal proxy Network Load Balancer serves TCP behind an internal IP reachable only from the same VPC or hybrid-connected clients, and publishing to other VPCs requires Private Service Connect.
| Option | IP advertisement | Backend placement | Availability behavior |
|---|---|---|---|
| Global with Premium Tier | Global anycast IP from all points of presence | Backends in multiple regions behind one backend service | Google Front Ends send traffic to healthy backends closest to the user; overflow goes to another region with capacity |
| Regional with Standard Tier or regional internal | Regional IP from points of presence for that region | Backends only in the same region as the forwarding rule | Traffic goes only to healthy backends in that one region |
Layer 4 proxies terminate TCP or SSL at a target proxy and open new connections to backends, flowing from forwarding rule to target proxy to backend service to backend. Layer 7 routing adds URL-map decisions, supporting cross-project referencing where a frontend and URL map in one service project reference a backend service in another. For hybrid NEGs with on-premises IP:Port endpoints, the VPC depends on Cloud Interconnect or Cloud VPN with Cloud Router using global dynamic routing over BGP, because regional routing and static routes are not supported.
Traffic distribution for external proxy Network Load Balancers is controlled by balancing mode on each backend. With CONNECTION mode, load spreads by total connections each backend can handle. With UTILIZATION mode, load spreads by instance utilization and applies only to VM instance-group backends. The classic proxy selects the most favorable group or NEG by balancing mode, then distributes in round robin within the group with consistent hashing within a zone. The global external proxy uses balancing mode for weight and traffic fraction plus a locality policy for sharing within the group.
Shared VPC separates network administration in a host project from service development in service projects. Regional external proxy and classic proxy Network Load Balancers support Shared VPC, while global external proxy does not. In supported deployments, the external IP must be in the same project as the load balancer, while the forwarding rule, target proxy, backend service, instance groups or NEGs, and health checks must be in the same project as the backend instances.
Private access and secure egress let VMs with only internal IPs reach Google APIs and the internet without direct exposure. Traffic from a private VM stays inside the VPC boundary and exits only through a controlled private endpoint or managed gateway. Private Google Access is a per-subnet setting that lets internal-only VMs reach supported Google APIs over the VPC internal path.
Cloud NAT is a managed egress service that translates a private source address to a shared public address for internet destinations and returns responses, without assigning an external IP to each VM. Deploy Cloud NAT in the Shared VPC host network so development, testing, and production share one controlled egress point. Private Service Connect (PSC) exposes Google APIs and cross-VPC services as internal endpoints: a PSC endpoint serves a bundle of global Google APIs, while a PSC backend serves a single API with load-balancer security features.
PSC between VPC networks uses a producer-to-consumer model where the producer publishes a service and the consumer creates an endpoint in its own VPC that forwards to the producer. The VPCs can be in the same or different organizations. When forwarding rules are used, consumer source IP is translated by source NAT to a PSC subnet address, so enable connection-information retention when the producer must see the original client IP. For centralized API access, direct routing places PSC endpoints in each VPC to bypass inspection devices, while indirect routing forces traffic through network virtual appliances (NVAs) in routing and transit VPCs for logging.
Shared VPC is a topology where one host project owns the VPC and subnets and service projects attach workloads with the Compute Network User role. Central teams control routes, firewalls, and IP allocation while service owners deploy VMs, Kubernetes pods, and serverless resources. A Cloud SQL instance can be created with a private IP in Shared VPC, but an existing instance cannot later gain a Shared VPC private address. Serverless VPC Access connectors let serverless workloads use internal IPs in VPC, peered, and hybrid-connected networks.
VPC Network Peering is a direct two-way connection for internal routing between two VPCs, and private services access uses peering internally to connect a customer VPC to Google-managed networks such as Cloud SQL. Peering is not transitive, so a third VPC cannot reuse another VPC's private services peering. To connect one private-IP Cloud SQL instance to multiple VPCs, use custom advertised routes, an intermediate SOCKS5 proxy with access to the peered VPC, or Cloud SQL Auth Proxy running inside the peered VPC with -ip_address_types=PRIVATE.
Hybrid connectivity extends VPCs to adjacent services through Dedicated Interconnect, Partner Interconnect, and HA VPN. Dedicated and Partner VLAN attachments support dual-stack IPv4 and IPv6, HA VPN supports dual-stack and IPv6-only gateways, and Classic VPN has no IPv6 support. Cloud DNS with DNS64 and Cloud NAT with NAT64 bridge IPv6-only clients to IPv4 destinations. Serverless VPC Access connectors route TCP and UDP, support ICMP only for external IPs, and do not support IPv6.
Exam tip: VPC Network Peering is not transitive, so a third VPC cannot reuse another VPC's private services access peering to Cloud SQL; use custom advertised routes, a SOCKS5 proxy, or Cloud SQL Auth Proxy as a service.
Dedicated Interconnect is a direct physical connection between on-premises and Virtual Private Cloud (VPC) networks without traversing the public internet or a provider, while Partner Interconnect flows through a supported service provider. Dedicated Interconnect requires installing routing equipment in a Google-supported colocation facility and uses 10-Gbps, 100-Gbps, or 400-Gbps circuits with up to 8 circuits per connection, while Partner Interconnect needs no colocation equipment and uses 50 Mbps to 50 Gbps VLAN attachments. Both provide access to Google Cloud products except Google Workspace and support Private Google Access to APIs from on-premises.
The 99.99 percent monthly uptime configuration requires four Cloud Interconnect connections across two Google Cloud regions in different metropolitan areas and separate edge availability domains. Valid compositions are four Dedicated Interconnect connections, four Partner Interconnect VLAN attachments, or two Dedicated Interconnect connections in one region plus two Partner Interconnect attachments in another region. A single Dedicated Interconnect connection and a single Partner Interconnect VLAN attachment both carry No Uptime SLA and do not qualify.
Virtual Private Cloud (VPC) Network Peering, Shared VPC and Private Service Connect differ by connectivity model, cross-organization support and isolation boundary. VPC Network Peering exchanges routes between two VPC networks for virtual machine to virtual machine reachability across projects and organizations, keeps separate firewall and Identity and Access Management (IAM) policies, requires non-overlapping IP addresses and is not transitive. Shared VPC lets service projects share subnets in one host-project VPC inside one organization with central network control, while Private Service Connect lets a consumer VPC connect privately to a producer service without merging networks within or across organizations.
Layer 3 and Layer 4 volumetric and protocol attack protection is always on for protected load balancers with automatic inline detection and mitigation by Google. Layer 7 application and HTTP-flood protection requires a user-configured Google Cloud Armor security policy attached to a backend service with custom filtering and Web Application Firewall (WAF) rules. Baseline protection automatically covers global external, classic, and external proxy Network Load Balancers for HTTP, HTTPS, HTTP/2, and QUIC.
Last updated