Certifications · Cisco

CCDE Practical: AI Infrastructure (CCDE AI)

A study series for the AI Infrastructure area of expertise of the CCDE v3.1 Practical, following Cisco's public technology list: requirements, options and trade-offs, decisions and their justification, with original practice scenarios.

1 of 19 articles published

The exam

Format

8 hours: four scenarios of up to 2 hours each

AI module

The fourth scenario, in the area of expertise you select (AI Infrastructure)

Navigation

No going back: an answer is final once you move on

Scoring

A minimum score per scenario and an overall pass score

Prerequisite

CCDE Written (400-007)

Materials

Closed book, with access to relevant cisco.com documentation; dual-stack throughout

From the official Cisco page, checked October 2026. Always confirm the current details there before booking.

How the module is weighted

Cisco publishes no weights for the AI Infrastructure technology list. The module is one scenario of four, and the core exam topics (business strategy, control and data planes, network, service and security design) apply to it as well.

Roadmap

The exam and the method

  1. 1

    How the AI Infrastructure module works, and a method for design scenarios

    Format and scoring, the technology list, what "design" means here, a method from requirements to justification, a worked mini-scenario

    Published

AI/ML, compliance and governance

  1. 2

    From use case to infrastructure: training, fine-tuning and inference

    ML, deep learning, LLMs and SLMs, and what each asks of compute, network and storage

    Coming · Nov 27, 2026
  2. 3

    Placement, data sovereignty, governance and lock-in

    On-premises, cloud or distributed; sovereignty and compliance; moving workloads between sites; multi-cloud and interoperability

    Coming · Dec 1, 2026
  3. 4

    Sustainability, power, cooling and cost

    Green AI, accelerator utilization, power and cooling as hard constraints, cost trade-offs

    Coming · Dec 4, 2026

Network

  1. 5

    Ethernet or InfiniBand: making and defending the choice

    Requirements that decide the transport, operations and skills, integration, the evidence to ask for

    Coming · Dec 8, 2026
  2. 6

    Lossless fabrics, QoS and RDMA as design decisions

    When lossless is required, PFC and ECN trade-offs, QoS consistency, RoCEv2 at design level

    Coming · Dec 11, 2026
  3. 7

    Bandwidth, latency, optics and cables

    Capacity planning, oversubscription, where latency comes from, optics versus direct-attach copper

    Coming · Dec 15, 2026
  4. 8

    Connectivity models: islands, the main data center, Clos and rails

    Purpose-built islands or the main data center, two- and three-tier Clos, rail-optimized designs

    Coming · Dec 18, 2026
  5. 9

    Routing and Layer 2 in AI fabrics

    Use-case optimizations for BGP and IGPs, ECMP, Layer 2 scale and reachability learning, EVPN

    Coming · Dec 22, 2026
  6. 10

    Traditional or SDN fabrics, and integrating with the rest of the network

    Controller-based fabrics such as ACI, interconnecting AI fabrics with the existing data center and WAN

    Coming · Dec 29, 2026
  7. 11

    Resiliency, DR and single- versus multi-site

    Failure domains, checkpointing and recovery, what multi-site means for training and for inference

    Coming · Jan 5, 2027

Security

  1. 12

    Hosting implications and infrastructure protection

    Air-gapped, public and private hosting, protection against malicious use, web application firewalls

    Coming · Jan 8, 2027
  2. 13

    Resource security: on-premises, cloud, clusters and containers

    Segmentation and multi-tenancy, shared GPU clusters, Kubernetes, cloud responsibility boundaries

    Coming · Jan 12, 2027

Hardware and environment

  1. 14

    Compute and AI-enabling hardware: GPU, SmartNIC, DPU

    What each device changes in the design, scale-up versus scale-out, offloads

    Coming · Jan 15, 2027
  2. 15

    Storage for AI

    Data storage strategy, FC, FCoE, NVMe and NVMe-oF, IP storage, separate or converged fabrics

    Coming · Jan 19, 2027
  3. 16

    Operations, telemetry and timing

    Telemetry analysis, log correlation, data scrubbing, PTP and why time matters

    Coming · Jan 22, 2027

Practice

  1. 17

    Practice scenario: an enterprise fine-tuning and inference platform

    A full scenario with changing requirements, answered and justified

    Coming · Jan 26, 2027
  2. 18

    Practice scenario: multi-site inference under sovereignty rules

    Placement, transport, security and DR under regulatory constraints

    Coming · Jan 29, 2027
  3. 19

    Exam-day method and final checklist

    Time per scenario, reading for the deciding requirement, the traps to avoid

    Coming · Feb 2, 2027
Built from Cisco's public exam topics, exam format and technology lists. The practice scenarios are my own; nothing here comes from the exam, and I'm not affiliated with Cisco. New articles arrive here first; follow the RSS feed to get them as they're published.