Networking for Customer Engineers
Reason about addressing, routing, enterprise connectivity, and operational evidence across cloud boundaries.
Addresses, subnets, CIDR, and packet boundaries
Interpret CIDR ranges
Routing, BGP, and anycast without magical thinking
Separate route learning from packet forwarding
Load balancing, NAT, and segmentation
Separate load balancing, NAT, and segmentation
VPN, MPLS, SD-WAN, and SASE as architecture choices
Explain enterprise connectivity layers
Multi-cloud connectivity and network observability
Trace a specific flow across clouds
Diagnose enterprise connectivity failure
A branch can reach the public Internet but cannot reach a private application after a network migration. The branch uses 10.20.0.0/16; a cloud network also uses 10.20.0.0/16.
Work through the problem
- Draw forward and return paths.
- List DNS, overlap, routing, firewall, and MTU hypotheses.
- Use route-table evidence to test the overlapping-address hypothesis.
- Propose a short-term scoped mitigation and a durable addressing plan.
Deliverables
- Evidence-led diagnosis
- Corrected network diagram
- Migration and verification plan
Self-review rubric
- Identifies overlapping prefixes
- Checks return routing
- Does not equate ping with application health
- States ownership and rollback
Compare with the worked approach
The duplicate 10.20.0.0/16 makes destination interpretation ambiguous across the connected networks. Confirm the selected route and expected destination before changing policy. A durable solution is coordinated non-overlapping addressing; controlled translation or routing isolation may provide a limited interim design. Retest DNS, bidirectional transport, application access, and unaffected sites.