Cisco CCDE

The best technical design can still be the wrong one. You weigh capital against operating cost, estimate return, account for skills and lifecycle, and show how governance and risk appetite change what is acceptable.

• Capital and operating cost over a lifecycle
• Estimating return and payback
• Risk appetite and risk transfer
• Skills, support and vendor dependency
• Governance and design approval processes
• Presenting a trade-off to a non-technical sponsor

You build A cost and risk comparison of two designs, with a recommendation a sponsor can accept.

Control plane choices decide how a network scales and how it fails. You select and combine IGPs and BGP, set failure domains and summarization boundaries, and predict convergence before anything is built.

• Choosing between link-state and path-vector routing
• IGP design: areas, levels and scaling limits
• BGP design: peering, reflection and policy
• Summarization and failure-domain boundaries
• Convergence behavior and its limits
• Redistribution and route-leaking hazards
• Predicting what breaks when a link fails

You build A routing design for a multi-region network with convergence and scaling arguments.

Packets must arrive in order, on time and at the right size. You design overlays and underlays, plan MTU, load sharing and multicast, and place QoS so that real traffic gets the treatment the business expects.

• Underlay and overlay interactions
• MTU, fragmentation and overlay overhead
• Equal-cost load sharing and polarization
• Multicast forwarding design
• QoS design across domains
• Asymmetric paths and stateful devices

You build A forwarding design note that resolves MTU, load-sharing and QoS issues for a given topology.

A network that cannot be operated is a failed design. You design out-of-band access, automation and orchestration, observability and assurance, and the measures that tell you how users and applications actually experience the network.

• Out-of-band and in-band management design
• Automation and orchestration architecture
• Telemetry and observability design
• Assurance against service-level objectives
• Measuring user and application experience
• Change management and rollback in design
• Operating model and team skills

You build An operations design covering management access, automation, telemetry and assurance targets.

Networks stay manageable when they are built from repeatable modules. You apply hierarchy, modularity and fault isolation to campus, WAN and data center, and decide where scale demands a new module rather than a bigger box.

• Hierarchy and modularity as scaling tools
• Fault isolation and blast radius
• Campus, WAN and data center module patterns
• Choosing Layer 2 or Layer 3 at the edge
• Growth planning and the point of redesign
• Evaluating a design against its failure scenarios

You build A modular design for a growing organization, with a growth path and its break points.

Software-defined options solve some problems and create others. You compare SD-WAN, campus fabrics and data center fabrics against traditional designs, and decide when a controller’s benefits outweigh its dependencies.

• What software-defined approaches change
• SD-WAN against traditional WAN design
• Campus fabric against classic campus
• Data center fabric options
• Controller placement, failure and dependency
• Migration paths from the existing network

You build A decision paper that selects an architecture for a multi-site organization and plans its migration.

Resilience is a requirement with a price. You choose redundancy models, design for the failures that actually happen, plan recovery, and show where resilience stops being worth its cost.

• Availability targets and what they imply
• Redundancy models and their hidden single points
• Fast failure detection and fast recovery
• Common-mode failures and shared fate
• Disaster recovery and site failover design
• Testing a design against failure

You build A failure analysis that finds the shared-fate risks in a sample design and fixes them.

AI clusters turn familiar networking assumptions upside down. You study the bandwidth, latency and loss behavior of training and inference traffic, and design the fabrics, storage paths and operations that keep expensive compute busy.

• Training and inference traffic patterns
• Back-end fabrics and non-blocking design
• Lossless Ethernet and its design risks
• Storage and data pipeline connectivity
• Scaling a cluster and its network together
• Operating and observing an AI network

You build A network design for a GPU cluster, with scaling limits and failure behavior described.

Services run across clouds, sites and edge locations, and the network joins them. You design cloud and hybrid connectivity, internet and SaaS access, and the placement of AI and machine learning services relative to their data.

• Cloud on-ramps and private connectivity
• Multi-cloud and data center interconnect
• Internet breakout and SaaS access design
• Where to place AI and ML services
• Latency, data gravity and egress cost
• Service resilience across providers

You build A hybrid service design that places workloads, links and breakout points with reasons.

Law shapes architecture. You translate data protection, residency and industry rules into design constraints, show how a design demonstrates compliance, and handle the extra duties AI services bring.

• Reading a regulation as a design constraint
• Data residency and sovereignty
• Industry rules such as payment-card requirements
• Evidence: logging, audit and retention
• Compliance for AI and ML services
• Cross-border traffic and provider choice

You build A compliance mapping that ties each rule in a scenario to a design decision.

Security must be built into the architecture, not attached to it. You place controls for segmentation, access and inspection, tie them to visibility and assurance, and revise policy for threats and tools that AI changes.

• Zero trust and segmentation in design
• Perimeter, edge and cloud security placement
• Visibility and assurance requirements for security
• Encryption and key management choices
• Integrating security with automation
• AI-enabled threats and AI use policy

You build A security architecture for a hybrid organization with controls justified against threats.

Expert design is defended, not just drawn. You take a full scenario, produce a design across all areas, then adapt it as requirements change and explain what each change costs.

• Analyzing a full scenario
• Producing a design across strategy, planes, services and security
• Adapting the design to a changed requirement
• Documenting trade-offs and rejected options
• Presenting and defending the design

You build A complete design document for a scenario, with an amendment for a changed requirement.