Cloud & Infrastructure Roadmap

Nine levels from deriving infrastructure out of a workload, through compute, networking, identity and delivery, to reliability, cost and the judgement to refuse complexity the business never asked for.

0 / 126 roadmap lessons mastered0%
  1. 1

    Compute, storage, networking, regions

    0 / 11 mastered

    Derive infrastructure from a workload instead of from a product catalogue.

    What Infrastructure Actually IsWorkload-First ThinkingFrom Laptop to ProductionOn-Premises vs CloudRegions and Availability ZonesThe Infrastructure Layer StackShared ResponsibilityCompute FundamentalsCompute as a Resource EnvelopeObject, Block and FileCloud Networking: Composing the Fundamentals
    Outcome · You can list what an application needs before naming a single service, and you can say where the provider's responsibility ends.
    Prove it — From Laptop to Production →
  2. 2

    VMs, containers, load balancing, DNS

    0 / 11 mastered

    Run a workload on something other than your laptop, and let traffic find it.

    Virtual MachinesHypervisors and Shared HostsThe VM LifecycleMutable Servers and Immutable ImagesOne Big VM or Several Small OnesThe Instance LifecycleContainers in OperationContainers vs Virtual MachinesLoad Balancers as InfrastructureDNS in Cloud InfrastructureCDN as Infrastructure
    Outcome · You can explain what a VM actually is, why a container is not one, and how a request reaches either.
  3. 3

    Storage types, managed databases, IAM

    0 / 11 mastered

    Put state somewhere defensible, and give every workload its own identity.

    Object StorageDirect Uploads and Signed AuthorizationBlock StorageFile StorageChoosing a Storage TypeManaged DatabasesManaged vs Self-HostedThe IAM ModelHuman vs Workload IdentityAnatomy of a PolicyLeast Privilege in Infrastructure
    Outcome · You can pick a storage shape from the access pattern, and write a policy that would survive a compromise.
    Prove it — IAM debugging lab →
  4. 4

    Docker, registries, CI/CD

    0 / 10 mastered

    Make the thing you tested the thing that runs.

    What Is Inside a Container ImageDocker Fundamentals — One Implementation of the ModelThe Container Build PipelineWhy Image Size Is an Infrastructure ProblemThe Container RegistryConfiguration Belongs Outside the ImagePersistent Data and ContainersThe Pipeline as InfrastructureBuild Once, Promote the Same BytesThe Infrastructure Supply Chain
    Outcome · You can trace an artifact from a commit to the digest production actually pulled, and name who was allowed to do each step.
    Prove it — Follow a Deployment →
  5. 5

    Infrastructure as code

    0 / 8 mastered

    Stop changing production by hand.

    Infrastructure as CodeDeclarative vs Imperative InfrastructureTerraform: The Vocabulary of Declarative InfrastructureState: The File That Makes It Work and the File That Will Hurt YouReading a Plan Before You Apply ItDrift: When the File and Reality DisagreeModules: Reuse Without HidingDevelopment, Staging and Production
    Outcome · You can read a plan, explain what state is for, and say what the next apply would do to a resource someone created by hand.
    Prove it — Drift lab →
  6. 6

    Networking boundaries, orchestration, autoscaling

    0 / 33 mastered

    Draw the trust boundaries, then decide whether you need a control plane at all.

    Virtual Private CloudPublic and Private SubnetsRoute TablesInternet GatewayNAT GatewaySecurity Groups: The Stateful FirewallNetwork ACLs: The Stateless FilterPrivate ConnectivityWhy Orchestration ExistsKubernetes: Why It ExistsThe Kubernetes Mental ModelCore Objects, and Why Each One ExistsPods: Shared Lifecycle, Shared NetworkDeployments and the Replica ControllerSelf-Healing, and What It Does Not HealService: A Stable Name in Front of Moving PodsIngress and Gateway: Getting Traffic InConfigMap vs Secret — and the Honest Limit of a SecretStateful Workloads: Databases Are Not Stateless APIsScheduling: How a Pod Chooses a NodeRequests vs Limits: Two Numbers That Do Different JobsOOM Kills and CPU ThrottlingHorizontal Pod Autoscaling — and Why New Capacity Is Always LateKubernetes Is Not Always NeededServerless as an Execution ModelServerless Trade-offsServerless and Database ConnectionsChoosing a Compute ModelAutoscalingAutoscaling SignalsStartup Time & Cold StartHealth ChecksLiveness vs Readiness
    Outcome · You can justify — or refuse — orchestration for a given workload, and you know why new capacity is always late.
    Prove it — Do we need Kubernetes? →
  7. 7

    Reliability, multi-zone, backup and recovery

    0 / 14 mastered

    Make redundancy real, and prove the restore works.

    Infrastructure ReliabilityHigh AvailabilityFailure DomainsMulti-Zone DeploymentDisaster RecoveryRPO & RTOBackup StrategyRestore TestingGraceful Shutdown: The 502 Spike Nobody InvestigatesFour Ways to Replace Running CodeRolling Deployment and the Compatibility It DemandsBlue/Green: Two Environments, One SwitchCanary: Let 5% of Traffic Find the BugDeployment Is Not Release
    Outcome · You can find the failure domain hiding behind three replicas, and you have restored a backup on purpose.
    Prove it — Break This Infrastructure →
  8. 8

    Multi-region, cloud security, cost engineering

    0 / 19 mastered

    Take on the expensive decisions with their costs stated out loud.

    Multi-Region DeploymentActive-Passive FailoverActive-ActiveThe Security ViewPublic Exposure, Read With ContextInfrastructure Trust BoundariesRoles vs Static KeysSecrets in InfrastructureKey Management and Encryption at RestInfrastructure ObservabilityInfrastructure LogsAudit TrailsCost EngineeringFixed vs Variable CostIdle Capacity: Headroom or Waste?Right-Sizing Without Causing an OutageCost per Service and the Attribution ProblemEgress: Moving Data Costs Money, Not Just Storing ItStorage Lifecycle: Hot, Warm, Archive, Delete
    Outcome · You can present an architecture in four views — structure, reliability, security and cost — and defend each.
    Prove it — Cost engineering →
  9. 9

    Production infrastructure design

    0 / 9 mastered

    Receive "we built an application, how do we run it?" and answer it systematically.

    Infrastructure for Model and Agent WorkloadsGPU and Accelerator InfrastructureHosted APIs, Managed Inference or Your Own ClusterMulti-Cloud, Taught CautiouslyHybrid CloudMigrating a System to the CloudThe Five DispositionsScoring Operational ComplexityNo Cargo-Cult Infrastructure
    Outcome · You can design infrastructure that meets the requirement and argue convincingly for everything you chose not to build.
    Prove it — Production SaaS capstone →