Knowledge Check - Networking and Connectivity
Test your understanding of the key concepts from Module 3. Select the best answer for each question, then submit to see your results. You need to score at least 60% to pass.
FinCorp's payment switch runs on two VMs and speaks a custom TCP protocol with mutual TLS. The risk office requires that the certificates stay on the servers, that no internet path to the switch exists, and that only the portal tier can reach it. Which design should the architect recommend?
The NLB forwards TCP without terminating TLS, so the certificates stay on the backends, and a private data LAN without an internet access element removes any internet path by topology. Because the NLB presents its backend address to the targets, the data-tier NSG allows that address rather than the clients. An ALB terminates TLS and takes no NSG, and the default NSG's broad rules do not express "portal tier only".
FinCorp attached a strict custom NSG to its data servers that allows only the database port from the NLB backend address. A flow log still shows ACCEPT records for SSH from another subnet in the VDC. The data servers were created after the VDC, which has a default NSG with its predefined rules. What is the most likely cause, and the fix?
Every new VM joins the default NSG, and a VM inherits the rules of every NSG it belongs to, so the predefined ingress rule still allows traffic that the custom group omits. NSGs are allow lists without a rule order, so reordering changes nothing. Running the NIC-based firewall in parallel with NSGs is advised against because the behaviours can conflict, and flow logs record what the network actually accepted or rejected.
FinCorp exposes a partner API on its public ALB. Only three partner networks may use the /partner/ path, while the customer portal on the same listener stays open to everyone. Which approach fits the ALB architecture?
The ALB is not a VM or NIC that an NSG can attach to, and its targets see the ALB's backend address, so a target NSG cannot distinguish partners; it should allow only the balancer. Client filtering therefore lives at Layer 7, as a source-IP condition on the partner path and as a check on the client address the ALB forwards. The NAT Gateway is SNAT only and never an inbound path.
A vendor appliance pair provides FinCorp's partner file gateway. The vendor's HA uses VRRP, and partners connect to one public address. The architect wants the address to move automatically when the active appliance fails. Which design works on IONOS CLOUD?
IP failover works on public LANs that contain no load balancer, with reserved IPv4 addresses only, and VRRP is supported based on gratuitous ARP once an IP failover group exists. The platform does not monitor the service; the HA software on the VMs detects the failure and announces the move. Virtual MAC addresses, and therefore CARP, are not supported, and IP failover is IPv4 only.
FinCorp's private servers in the Frankfurt VDC must replicate to the recovery VDC in Berlin over a private, encrypted link, and must download patches from the internet without any public interface. Both VDCs are in the same contract. Which combination should the architect choose?
Cross Connect requires the same region as well as the same contract, while the VPN Gateway connects VDCs across regions with static routes. The NAT Gateway connects only to private LANs, provides SNAT for outbound traffic, and does not inject its route, so each VM needs its default route set to the gateway. BGP is not supported, and a public LAN would give the servers the exposure the requirement forbids.
The risk office wants FinCorp's portal to fail over automatically from Frankfurt to Berlin, with clients following within a few minutes of a site failure. The portal tier is stateless. Which design delivers this?
Cross-site failover is composed: the balancers keep each site healthy internally, and your monitor moves the record between sites through the API, with the TTL bounding how long clients keep the old answer. The Cloud DNS SLA covers the name servers, not the endpoints, and the default TTL of 3600 seconds would leave clients on the dead site for up to an hour. An IP failover group cannot leave its LAN, and reserved addresses are region-bound.