Knowledge Check - Platform Architecture Overview
Test your understanding of the key concepts from Module 1. Select the best answer for each question, then submit to see your results. You need to score at least 60% to pass.
FinCorp is laying out its customer portal in one German VDC. The portal serves HTTPS to the internet, its application servers must call an external payment API, and the PostgreSQL database must never be reachable from the internet. Which layout best fits the canonical layered architecture?
The canonical shape puts the Layer 7 edge in front, keeps compute private, and leaves the data tier on a LAN with no internet path, while the SNAT-only Managed NAT Gateway gives the private tier outbound access without opening an inbound one. Relying on NIC firewalls on a shared public LAN makes topology depend on filtering, which protects only server NICs. Targets cannot initiate connections through the ALB, because it does not support SNAT, so the payment call needs its own egress path.
A FinCorp team runs a read-heavy reporting service on DBaaS for PostgreSQL. The database primary is saturated by repeated reads of account summaries that change rarely. The team proposes adding standby nodes to spread the reads. What should the architect recommend?
In DBaaS for PostgreSQL, standby nodes run in warm standby mode and exist for failover, so adding them buys availability rather than read capacity. The native substitution for read scaling is a cache for stable hot data plus connection pooling. Changing the replication mode affects durability and availability, not whether standbys answer reads, and the cross-cluster option assumes a managed replication path that the design should not rely on.
FinCorp's auditor requires that the services in the customer-facing compute tier fall within the scope of the ISO 27001 certificate on the basis of IT-Grundschutz. The current design runs that tier on Cubes and protects the VMs with Backup Service. What is the correct response?
The IT-Grundschutz certificate covers Compute Engine, Managed Kubernetes and IONOS Cloud Object Storage in Germany; Cubes and Backup Service are outside it, so the compliance filter forces a product change and an explicit record of the backup gap. A provider-level claim does not extend to every service, and a data-center credential describes the facility, not the service. Swapping in C5 does not answer an IT-Grundschutz requirement, and C5 scope must itself be confirmed per service in the attestation report.
To move fast, a FinCorp project team wants to build its platform in a VDC at a non-German location now, reserve its public IPv4 addresses there, and "move the VDC to Germany" once the compliance review is complete. How should the architect respond?
Residency is a placement decision: you specify a VDC location only at first provisioning, and the region-bound artifacts cannot follow a later change, so the compliance filter must run before workloads land. Cross Connect joins VDCs only within one region and contract and does not relocate anything. An ownership transfer changes the contract holder, not where the resources run.
FinCorp's API team asks for "an API gateway" in front of a set of microservices on Managed Kubernetes. They need routing by host and path, TLS termination at the edge, and the ability to add services without changing the edge for every release. Which composition should the architect propose?
Applying the native-substitution model, the coarse routing and TLS offloading sit in ALB forwarding rules, and the fine-grained per-service routing lives in an ingress controller inside the cluster, so new services change only the ingress rules. A Service of type LoadBalancer assigns a static IP to one node rather than provisioning a load balancer, so one per microservice multiplies single-node entry points. A Layer 4 balancer cannot route on host or path, and IP failover is an availability mechanism, not a router.