TELCOMA Certified Telco Cloud Specialist

What's assessed

7 domains, 49 questions in total — drawn from the course lessons.

• The Kubernetes control plane for telco: place, heal, scale

  7 questions

  Why a crashed container is back in seconds, a lost server's pods wait out the eviction timer, and a stateful replica waits for proof its old copy is gone; level-triggered reconciliation against one record; the workload kind each function needs; and the frozen fraction one wrong processor limit imposes on a poll-mode datapath.

• CNF engineering: state, probes and discovery

  7 questions

  The disposability contract and the user plane as its hard exception; what saving state after every message costs; stateless mobility-function failover and its three scale-out bugs; readiness as a claim, and a grace window counted from deletion; and the two discovery planes, cluster DNS and the network's own directory.

• The telco node: pinning, hugepages, NUMA and real time

  7 questions

  Why Guaranteed alone does not buy exclusive processors; large pages and the translation cache, and what a node with none set aside does; aligning processor, memory and card on one memory node; isolated cores, interrupts and timing for a DU; and the labels and tuning profiles that make hardware schedulable.

• The packet path: Multus, SR-IOV, DPDK and the UPF datapath

  7 questions

  Wiring a user-plane pod's reference points with addresses that survive a restart; how a virtual function reaches a pod and where its address is recorded; poll-mode drivers and the privilege their namespace needs; choosing a datapath mode on footprint rather than throughput bars; and what a service mesh costs the signalling path.

• Operators, admission, packaging and the management plane

  7 questions

  Teaching the API a new noun, and the order pod admission runs in; Operators against one-shot chart releases when a member is lost; where the NFV management plane sits above the cluster and what its linking specification does with the Kubernetes interface; where per-site variance lives; and what a conformance badge attests.

• Day-2 operations: disruption, upgrades, scaling and failure

  7 questions

  Disruption budgets that block a drain; StatefulSet rolling updates, the shutdown sequence and the revert that is not enough; a node-reboot upgrade campaign on the reference cluster; autoscaling on an honest signal; a lost majority that leaves pods running while nothing changes; and the traces, counters and probes that show it.

• GitOps, fleets and what is actually shipping

  7 questions

  Pull against push, and where drift becomes visible; self-heal fighting the autoscaler, and who owns a field; a reviewed change that ships, reconciles and reverts, and what a revert cannot undo; configuration owners and Secrets that are in sync and still wrong; declared clusters across an edge fleet; and which automation projects ship.