Study ANS-C01 implementation of Transit Gateway, VPC peering, PrivateLink, AWS RAM, route tables, security boundaries, DNS, and connectivity validation.
This ANS-C01 task is about turning a multi-account or multi-VPC design into working AWS routing without losing ownership, segmentation, DNS, or testability. The exam is not satisfied by “use Transit Gateway.” It expects you to know which account owns the hub, how attachments are shared, which route tables receive propagation, which VPC route tables point to the hub, and how the path is validated.
The current AWS task statement is to implement routing and connectivity across multiple AWS accounts, Regions, and VPCs for different connectivity patterns. The official scope includes inter-VPC and multi-account connectivity, PrivateLink, AWS Organizations, AWS RAM, DNS, infrastructure automation, authentication and authorization, network security controls, and validation tools such as Reachability Analyzer and Route Analyzer.
| Implementation question | What a strong answer includes |
|---|---|
| Who owns the shared network resource? | A hub or network account, AWS RAM sharing, and account-level permissions. |
| Which connectivity pattern is being implemented? | VPC peering, Transit Gateway, PrivateLink, VPN, SD-WAN, or third-party path matched to the requirement. |
| Which route domain should each attachment use? | Transit Gateway route table association, propagation, static routes, or blackhole routes where appropriate. |
| What changes inside each VPC? | Subnet route tables, security groups, NACLs, endpoint policies, and DNS settings. |
| What crosses the account boundary? | RAM shares, Organizations scope, IAM/SAML/Active Directory integration, and ownership of operations. |
| How is success proven? | Reachability Analyzer, Route Analyzer, flow logs, Network Manager, and service-specific health evidence. |
Implementation questions are often about sequencing. A design can be correct and still fail if the implementation omits route-table association, propagation, VPC subnet routes, endpoint permissions, or DNS.
| Requirement | Strongest first implementation |
|---|---|
| Many VPCs across accounts need centralized routing | Shared Transit Gateway with AWS RAM, explicit route tables, associations, and propagations. |
| A few VPCs need simple direct private connectivity | VPC peering plus routes in both VPCs and non-overlapping CIDRs. |
| One private application service should be consumed by many VPCs | PrivateLink endpoint service with consumer interface endpoints and endpoint permissions. |
| Multi-Region hub connectivity is required | Transit Gateway peering or Cloud WAN pattern with explicit regional route tables. |
| Existing SD-WAN or appliance overlay must connect to AWS | Transit Gateway Connect, VPN, or vendor integration matched to the overlay design. |
| Cross-account resources need centralized network ownership | AWS Organizations and AWS RAM, with clear resource-owner and participant-account responsibilities. |
| Connectivity must be repeatable at scale | Infrastructure automation plus validation, not console-only setup. |
Use this sequence when the stem describes a hub-and-spoke implementation.
flowchart LR
H["Network account owns TGW"] --> S["Share with accounts using AWS RAM"]
S --> A["Create VPC / VPN / DXGW / peering attachments"]
A --> T["Associate each attachment with one TGW route table"]
T --> P["Propagate routes into intended route tables"]
P --> V["Add VPC subnet routes to TGW"]
V --> X["Validate path and logs"]
The most common implementation miss is assuming that attaching a VPC to a Transit Gateway automatically updates every relevant route table. It does not. The Transit Gateway route table and the VPC subnet route table both need the correct routes for the intended direction.
| Term | Meaning | Exam trap |
|---|---|---|
| Attachment | The connection between TGW and a VPC, VPN, Direct Connect gateway, peering connection, or other supported target. | Creating an attachment does not by itself prove reachability. |
| Association | The single TGW route table an attachment uses to look up destinations for traffic entering the TGW from that attachment. | An attachment can be associated with only one TGW route table. |
| Propagation | Installing the attachment’s routes into one or more TGW route tables. | Propagation controls what other route domains can learn, not which table the source attachment uses. |
| Static route | Manually entered TGW route, often needed for peering or appliance patterns. | Some paths, including TGW peering, need static route thinking. |
| Blackhole route | Route that intentionally drops matching traffic. | Useful for guardrails, but dangerous if mistaken for a normal route. |
When a question says one spoke cannot reach shared services, ask both questions: which route table is the source attachment associated with, and which route table received the destination route?
AWS RAM is the normal implementation tool when a central account shares a Transit Gateway or other shareable network resource with participant accounts.
| Step | What to verify |
|---|---|
| Share the TGW from the network account | The share targets the right accounts, OU, or organization scope. |
| Accept or use the share from participant accounts | The attachment account can create the needed VPC attachment. |
| Keep route-table control clear | The network account commonly owns central TGW route table associations and propagation. |
| Update VPC route tables in spoke accounts | Subnets that should use the hub need routes to the TGW attachment. |
| Tag and log consistently | Multi-account operations need searchable ownership and evidence. |
The exam may test account ownership. A participant account can attach its VPC to a shared Transit Gateway, but central route-table control and governance usually remain in the network account.
PrivateLink is an implementation answer when consumers need private application access without broad network routing.
| Provider side | Consumer side |
|---|---|
| Put the service behind a supported load balancer or service resource. | Create an interface endpoint in the consumer VPC. |
| Create an endpoint service. | Select subnets and security groups for endpoint ENIs. |
| Allow specific principals or accounts to connect. | Accept or request endpoint connection depending on service settings. |
| Decide private DNS behavior. | Use endpoint DNS names or private DNS if configured and appropriate. |
| Monitor service and endpoint health. | Validate security group, route, DNS, and application reachability. |
PrivateLink is weak when the requirement is broad bidirectional routing. It is strong when the requirement is private access to a specific service across VPCs, accounts, or sometimes from on-premises through private connectivity.
VPC peering is still valid when the topology is small and direct.
| Check | Why |
|---|---|
| CIDRs do not overlap | Peering cannot solve ambiguous destination ranges. |
| Routes exist in both VPC route tables | Peering is not useful if only one direction is routed. |
| Security groups and NACLs allow the path | Routing success does not imply security success. |
| DNS behavior is explicitly configured where needed | Cross-VPC name resolution may require settings and private hosted zone planning. |
| No transitive routing is required | VPC peering does not turn one VPC into a transit hub. |
If the scenario says “many VPCs,” “central inspection,” “shared hybrid connection,” or “segmented route domains,” peering is usually a weaker implementation choice.
Task 2.2 explicitly includes host and service name resolution for applications and clients. Multi-account connectivity is incomplete if applications cannot resolve the right private names from the right accounts and VPCs.
| Need | Implementation cue |
|---|---|
| Shared internal names across VPCs | Private hosted zone association or shared-resolution architecture. |
| On-premises to AWS private DNS | Resolver inbound endpoint and on-premises conditional forwarding. |
| AWS to on-premises DNS | Resolver outbound endpoint and forwarding rules. |
| Private application endpoint exposed through PrivateLink | Endpoint DNS, private DNS name, and endpoint policy behavior. |
| Multi-account DNS consistency | Central DNS account, Resolver rule sharing, or delegated zone model. |
Do not mark a network implementation complete until name resolution and packet routing both work.
Connectivity at scale can accidentally flatten the network. ANS-C01 expects implementation choices that preserve boundaries.
| Boundary requirement | Strong implementation |
|---|---|
| Prod and nonprod must remain isolated | Separate TGW route tables, selective propagation, blackhole routes where useful, or Cloud WAN segments. |
| Shared services should be reachable but not everything else | Route only required prefixes and use security groups, NACLs, firewall policy, and endpoint controls. |
| East-west traffic must be inspected | Route through AWS Network Firewall or GWLB appliance path with symmetry. |
| Cross-account application access must be narrow | PrivateLink endpoint service permissions and endpoint policies. |
| Authentication or directory dependency crosses networks | Confirm SAML, Active Directory, DNS, and firewall requirements together. |
The strongest implementation answer usually limits reachability deliberately instead of relying on after-the-fact security review.
Implementation is not finished until the path is proven.
| Validation need | Strong tool or evidence |
|---|---|
| Static path analysis inside AWS | VPC Reachability Analyzer. |
| Transit Gateway route-table behavior | Transit Gateway Route Analyzer, route tables, associations, and propagations. |
| Flow accepted or rejected | VPC Flow Logs and Transit Gateway Flow Logs. |
| Hybrid or global topology health | AWS Network Manager and CloudWatch metrics. |
| Application service reached through PrivateLink | Endpoint status, DNS resolution, security group logs, service logs, and consumer request tests. |
| Multi-account rollout consistency | Infrastructure-as-code outputs, tags, config checks, and deployment pipeline evidence. |
If an answer only creates resources but never validates route behavior, it is incomplete for ANS-C01 implementation questions.
| Scenario | Strong implementation |
|---|---|
| Central network team owns shared hub; app accounts own VPCs | Share TGW with RAM, create VPC attachments, centralize TGW route-table control, update spoke route tables. |
| Production spokes can reach shared services but not nonproduction | Separate TGW route tables and controlled propagation. |
| Many consumers need one internal API | PrivateLink endpoint service with account permissions and endpoint DNS planning. |
| Two small VPCs need direct connectivity | VPC peering with reciprocal routes and security controls. |
| Inter-Region VPC fleet needs controlled routing | TGW peering with static route entries or Cloud WAN where global policy is central. |
| Existing SD-WAN appliance must integrate with AWS | Transit Gateway Connect or vendor-supported overlay implementation. |
| Stem clue | Eliminate first | Keep in play |
|---|---|---|
| “attachment created but traffic still fails” | assuming TGW attachment is enough | VPC route tables, TGW association, propagation, security, return path |
| “many VPCs and accounts” | full-mesh peering by reflex | RAM-shared Transit Gateway or Cloud WAN |
| “one application only” | broad routing to the provider VPC | PrivateLink |
| “prod and nonprod segmentation” | one flat TGW route table | separate route tables or segments |
| “shared service names fail to resolve” | route-table-only fix | private hosted zones, Resolver rules, DNS association |
| “inter-Region TGW path” | propagation-only assumption | TGW peering static routes and regional route tables |
| “prove before production cutover” | manual ping only | Reachability Analyzer, Route Analyzer, flow logs, metrics |
When a multi-account implementation does not work, use this order:
This order avoids the common mistake of changing firewall rules while the actual failure is a missing TGW propagation, a missing subnet route, or an unaccepted endpoint connection.
A central networking account owns an AWS Transit Gateway shared by AWS RAM with several application accounts. A new production VPC attachment was created successfully, but instances in that VPC cannot reach the shared services VPC. Other production VPCs can reach shared services. Which implementation checks are strongest first?
Answer: Verify TGW route-table association, propagation, VPC subnet routes, and return path.
Why: A successful attachment does not prove routing. The failure is limited to a new VPC, while other production VPCs work, so the first checks should be the new attachment’s route-domain placement, propagated routes, subnet routes toward the TGW, and return path. VPC peering is not needed, public exposure violates the private shared-services goal, and broad security changes are premature.
For ANS-C01 implementation, name the pattern first, then verify the implementation objects: share, attachment, association, propagation, subnet route, security boundary, DNS, and validation evidence. If the scenario is narrow service access, implement PrivateLink. If it is many VPCs and accounts, implement Transit Gateway or Cloud WAN with route-domain control. If it is only two simple VPCs, VPC peering can still be correct.
Continue with 2.3 Hybrid and Multi-Account DNS Implementation to implement the name-resolution layer that makes these paths usable by applications.