Foundations — Link-Local and SNI Quiz
This quiz tests your understanding of link-local/ULA addressing, how SNI works, and Lattice's protocol support.
Multiple Choice Questions
- Which statement about the address ranges a VPC Lattice service DNS name resolves to is correct?
- A) Both IPv4
169.254.171.0/24and IPv6fe80::/10are link-local - B) IPv4
169.254.171.0/24is link-local, while IPv6fd00:ec2:80::/64is not link-local but a Unique Local Address (ULA) - C) Both ranges are globally unique public addresses
- D) IPv6 is not supported
- A) Both IPv4
Show Answer
Answer: B) IPv4 169.254.171.0/24 is link-local, while IPv6 fd00:ec2:80::/64 is not link-local but a Unique Local Address (ULA)
Explanation: The IPv4 range falls inside 169.254.0.0/16 (RFC 3927, link-local), but the IPv6 range is not fe80::/10 (link-local) — it is fd00:ec2:80::/64 inside the fc00::/7 ULA range (RFC 4193). Link-local addresses have link scope. ULAs have global address scope under RFC 6724, but are intended for private routing rather than global Internet reachability; they are not the deprecated site-local address class. These address classifications do not by themselves describe AWS's special Lattice ingress implementation.
- How should the Lattice address ranges be interpreted?
- A) To conserve IP addresses
- B) Use AWS-documented service-specific addressing; do not turn the prefix into a general definition of link-local interception
- C) To cope with IPv4 address exhaustion
- D) To make the client explicitly aware of the destination
Show Answer
Answer: B) Use AWS-documented service-specific addressing; do not turn the prefix into a general definition of link-local interception
Explanation: The service-specific route and association determine reachability. ULA is not link-local or the deprecated site-local class; address scope and routing reachability differ.
- Lattice-bound requests appear in Envoy logs during coexistence. What should be checked next?
- A) Exceeding Lattice quotas
- B) Check whether mesh interception and the configured outbound route/signing policy explain the failure
- C) A Gateway API CRD version mismatch
- D) A Target Group protocol misconfiguration
Show Answer
Answer: B) Check whether mesh interception and the configured outbound route/signing policy explain the failure
Explanation: Some Envoy policies reject unknown destinations and others forward them. Inspect rules and logs, then test the chosen bypass or forwarding design; coexistence does not imply inevitable failure.
- Why is destination IP alone insufficient for stable service identity and authorization?
- A) Destination IPs become encrypted — there is no alternative
- B) Use service-aware identity, DNS/request correlation and reviewed network rules rather than relying on IP alone
- C) Flow logs get disabled — enabling them solves it
- D) The VPC CNI reuses IPs — changing the IP allocation policy solves it
Show Answer
Answer: B) Use service-aware identity, DNS/request correlation and reviewed network rules rather than relying on IP alone
Explanation: Service identity is not a stable per-service CIDR security boundary. Network evidence remains useful when combined with auth policies, service logs and request IDs. SG changes must use reviewed IaC.
- What is the correct SNI/ECH distinction?
- A) It is a leftover vulnerability from early TLS design
- B) Ordinary TLS exposes SNI to select a certificate; ECH is another design, but Lattice TLS passthrough does not support it
- C) SNI encryption was skipped for performance
- D) It was deliberately exposed to allow firewall traversal
Show Answer
Answer: B) Ordinary TLS exposes SNI to select a certificate; ECH is another design, but Lattice TLS passthrough does not support it
Explanation: TLS passthrough requires a custom domain matching SNI. AWS explicitly excludes ECH/ESNI there, so do not state that plaintext SNI is universally unavoidable or that Lattice support is merely unknown.
- Which statement about raw TCP in VPC Lattice is correct?
- A) TCP is not a supported protocol on the AWS network
- B) Raw TCP is absent from service listeners, but TCP resource configurations/resource gateways are a separate supported model
- C) An NLB already does this, so it would be redundant
- D) Security regulations prohibit plaintext communication
Show Answer
Answer: B) Raw TCP is absent from service listeners, but TCP resource configurations/resource gateways are a separate supported model
Explanation: Do not call this a product-wide mathematical impossibility. Compare the separate resource-connectivity access model, controller support and routing needs; do not classify h2c gRPC as arbitrary raw TCP.
- Which statement correctly describes the trade-off between an HTTPS listener and TLS Passthrough?
- A) TLS Passthrough is superior in every respect
- B) Passthrough preserves endpoint TLS but cannot inspect HTTP SigV4 identity; anonymous network-context policies are a different control
- C) Both listener types authenticate encrypted HTTP SigV4 headers without decryption
- D) An HTTPS listener cannot see SNI
Show Answer
Answer: B) Passthrough preserves endpoint TLS but cannot inspect HTTP SigV4 identity; anonymous network-context policies are a different control
Explanation: The loss is authenticated HTTP identity and L7 inspection, not every form of auth policy. Select listener and endpoint controls explicitly; backend HTTPS encryption does not mean Lattice validates target certificates.