Skip to content

Expert Networking Study Path

Resources checked: September 15, 2026

Environment: Disposable Ubuntu Server 24.04 LTS or Rocky Linux 9 VMs for host exercises. Routing/EVPN tools and cloud environments have separate chapter prerequisites.

These advanced workbooks follow the beginner course. Connect protocol principles to code and packets, then evaluate routing policy, Linux performance and cloud designs using evidence.

The expert-level objective is to define what should be true, test hypotheses with observations, and explain a change's impact and recovery. A fixed period of study or a certification alone does not guarantee that ability.

Entry check

Proceed to chapter 1 when you can explain and reproduce the following:

  • Distinguish IP, subnet, gateway and DNS roles from MAC/ARP.
  • Read ip route get, ss and a bounded tcpdump capture, identifying the observation point.
  • Verify SSH key access and the lab firewall policy, then restore the previous state.
  • Distinguish an HTTP error, connection failure and name-resolution failure.
  • Separate management and experiment paths and identify your own files, VMs and containers.

Use the beginner capstone to revisit missing skills. Complete guest tool preparation when a host command is unavailable.

Six workbooks

Each row follows reading → experiment or design validation → evidence → explained quiz. External courses supply reading scope; these workbooks supply a way to assess your own results.

OrderWorkbookDeliverable
1Protocol and implementation projects · QuizRequest path, probe/response correlation, protocol boundaries and edge cases
2Routing policy and convergence · QuizIntended advertisements/selection, RIB/FIB and forwarding evidence, failure/recovery timeline
3Data-center EVPN/VXLAN · QuizUnderlay/overlay mapping, allowed/forbidden VRF flows, routes and encapsulation evidence
4Linux packet paths and performance · QuizTime-correlated queues, counters and packets; measurement limits and recovery
5Cloud and CNI design · QuizDesign and evidence from DNS to backend and return path
6Automation and final assessment · QuizEvidence and change review covering normal, forbidden, failed and recovered states

Existing references remain useful: protocol foundations, kernel networking, Calico BGP, Cilium, and VPC CNI.

Choose a specialization

DirectionPriority orderFurther depth
Cloud networking/SRE1 → 2 → 4 → 5 → 6Performance and failures across CNI, DNS, load balancers and hybrid boundaries
Enterprise/data center1 → 2 → 4 → 3 → 6Check remaining L2/L3, security, virtualization and automation topics against Cisco's official scope
Protocol/network software1 → 2 → 4 → 6Transport and kernel implementation projects after establishing C/C++ foundations

Chapters 1, 2 and 4 are common prerequisites. Add chapter 3 for the data-center path or chapter 5 for cloud. The protocol/Linux path can use chapter 4's host experiment for the corresponding chapter 6 assessment. Complete the assessment in the selected scope, then add another specialization. Allocate time for environment preparation, failed attempts, reproduction and reporting as well as reading.

Lab boundaries and ownership

Do not execute every chapter sequentially in one environment.

EnvironmentPreparation and boundaryCheck before completion
Host observationDisposable beginner-course VM, installed tools, your own trafficInterface/process/capture scope, duration, cleanup of your files
Routing/EVPNSeparate Linux lab VM and selected netlab/provider installation/image requirementsUpstream revision, tool/image versions, owned lab ID/resources and recovery commands
CloudApproved existing lab account/cluster and read permissionsAccount, region, context, namespace, resource owner, cost/permission/validation scope
Design/supplied evidenceStated assumptions and input evidence without necessarily having the equipmentDistinguish design review from live traffic validation and mark unobserved items

Do not copy a netlab/containerlab device's addresses or names onto a beginner VM's NIC. Some exercises require commercial NOS images, virtualization support or additional boards: check each chapter and upstream prerequisites first. Global Docker prune, teardown of other labs and production failure injection are not default steps in this course.

Create this record before every experiment. Recording versions and revisions means capturing the actual inputs; it does not guarantee compatibility between arbitrary combinations.

RecordExample form
Experiment ID and ownerAn ID you can distinguish and the responsible role
Environment/permissionLocal VM, lab container, or approved cloud read-only access
TopologyNodes, interfaces, addresses, VRFs, observation and management paths
Tools/inputsOS, kernel, tool and image versions; upstream commit; input-file hashes
Traffic scopeSource, destination, protocol, ports and maximum duration
BaselineRoutes, policy, service state and recovery configuration
Hypothesis/changeExpected outcome and the single item to change
Stop/recovery criteriaManagement failure, unexpected resources, restoration and recheck

Selecting official resources

These are learning directions checked on September 15, 2026. Confirm access to semester materials, enrollment, graders and scheduled training separately.

ResourceRole in this courseBefore using it
Berkeley CS168Structure for Internet architecture, routing, transport, data centers and projectsSelect a populated public semester and check access to descriptions, code and grading separately
BGP LabsProgressive BGP configuration, policy and failure practiceCheck installation, lab revision and provider/image requirements
netlab EVPNEVPN/VRF/IRB design and feature-support validationVerify the selected devices support the required feature and data-plane setup
Bootlin Linux NetworkingKernel, driver, user-space and performance depthSeparate public material from paid training; check C/kernel and board requirements
Cisco Enterprise InfrastructureCheck enterprise design, operations and automation coverageSelect the current official blueprint; FRR does not reproduce every commercial device or ASIC behavior
AWS Advanced NetworkingScope for cloud/hybrid design, implementation, operations and securitySeparate exam preparation from deployment; establish account, access, cost and cleanup conditions
Stanford CS144Optional C/C++ direction for protocol implementationVerify current enrollment, public assignments and repository availability

This repository does not supply private course materials or solutions. Read the project requirements and produce your own implementation, tests and design justification.

Interpreting evidence

  • Measured: Observed in your stated environment, time window and traffic.
  • Supplied: Collected by someone else; record provenance, scope and relation to the original.
  • Illustrative/synthetic: Inputs explaining a calculation or validator. Do not report them as successful real traffic.
  • Design assumption: An untested design condition. Keep it marked unverified.

Absent observations are unknown. Empty routing output, missing packets and a ping timeout alone do not establish policy denial or an end-to-end root cause. An established BGP session does not establish the intended prefixes or forwarding behavior.

Final portfolio

Connect these deliverables in your own environment:

  1. Forward and return paths for one request, including observation points.
  2. Evidence for both normal traffic and traffic that must be forbidden.
  3. Before/during/after records for one fault or policy change.
  4. Measurement conditions, missing evidence and rejected hypotheses.
  5. A narrow change and post-recovery verification.
  6. Final-assessment automation results and their human-review limits.

Keep original evidence in an approved location and use consistent aliases in public notes. Do not publish raw packets, logs, accounts or topology inventories indiscriminately.

Next: Chapter 1 — Protocol and implementation projects