Cloud Pro: Azure Compute: Choose, Compare and Launch

What's inside

1. Level 1: The Azure Compute Map on One PageFree
   Learners map every Azure compute service on one page, grouped by hosting model (IaaS, managed containers, PaaS and serverless, specialised hosting), and line each one up with Microsoft's compute decision tree and its current and former names.
2. Level 2: The Standard Workload ProfileFree
   Learners record any workload in a standard ten-part profile covering compute type and memory, parallelism, duration and state, latency, scaling, OS and licensing, data location, compliance, team skills and cost model, which every later lesson uses.
3. Level 3: Virtual Machines: General Purpose Series
   Learners cover Azure Virtual Machines in the nine-part pattern, using the general purpose B, D and Dsv-type series, and deploy then delete the smallest working Linux VM with the Azure CLI and Bicep.
4. Level 4: Compute and Memory Optimised VMs
   Learners choose between compute optimised (F-series) and memory optimised (E, M-series) VMs by vCPU-to-memory ratio, and find out where licensing, quotas and regional availability change the decision.
5. Level 5: HPC VMs: The H-Series and InfiniBand
   Learners cover the HPC H-series (HB, HC, HX) in the nine-part pattern, including InfiniBand RDMA networking, HPC images and the placement and quota limits that decide whether a tightly coupled MPI job scales.
6. Level 6: GPU VMs: The N-Series
   Learners cover the GPU N-series (NC, ND, NV and their generations) in the nine-part pattern, matching GPU model, GPU memory and interconnect to training, inference, visualisation or GPU-accelerated simulation.
7. Level 7: Spot VMs and Dedicated Host
   Learners cover two ways of buying capacity other than standard on-demand: Spot VMs for interruptible work at deep discounts, and Dedicated Host for single-tenant physical servers that meet compliance or licensing needs.
8. Level 8: Virtual Machine Scale Sets
   Learners cover VM Scale Sets in the nine-part pattern, comparing Flexible and Uniform orchestration, autoscale rules, mixed Spot and standard instances, and upgrade policies for fleets of identical VMs.
9. Level 9: Azure Batch
   Learners cover Azure Batch in the nine-part pattern as a managed job scheduler for large-scale parallel and HPC work, with pools, jobs and tasks, autoscale formulas, Spot nodes and the account quotas that bite.
10. Level 10: Azure CycleCloud for HPC Clusters
    Learners cover Azure CycleCloud, including CycleCloud Workspace for Slurm, in the nine-part pattern for running and autoscaling traditional HPC schedulers (Slurm, PBS Pro, LSF and others) on Azure VMs.
11. Level 11: Azure Kubernetes Service
    Learners cover AKS in the nine-part pattern, including the control plane tiers, node pools (system, user, Spot and GPU), the AKS Automatic and Standard modes, cluster autoscaler, upgrades and the limits that bite.
12. Level 12: Azure Container Apps, Jobs and GPU Profiles
    Learners cover Azure Container Apps in the nine-part pattern, including the Consumption and Dedicated workload profiles, event-driven scaling to zero, Container Apps jobs for batch work, and serverless GPU workload profiles.
13. Level 13: Azure Container Instances
    Learners cover Azure Container Instances in the nine-part pattern as the simplest way to run a container or container group on demand without a cluster, including its per-second billing and the size and feature limits that bite.
14. Level 14: Azure Red Hat OpenShift
    Learners cover Azure Red Hat OpenShift in the nine-part pattern as a jointly operated, fully managed OpenShift service, comparing it with AKS on cost, support model, minimum cluster size and team skills.
15. Level 15: Azure App Service
    Learners cover App Service in the nine-part pattern for web apps and APIs on Windows or Linux, including plan tiers, deployment slots, scale-out limits, App Service Environment for isolation, and containers on App Service.
16. Level 16: Azure Functions: Flex Consumption and Other Plans
    Learners cover Azure Functions in the nine-part pattern, comparing the Flex Consumption, Premium, Dedicated and container-hosted options, and the legacy Consumption plan with its retirement status, by cold start, timeout, networking, scaling and cost.
17. Level 17: Azure VMware Solution
    Learners cover Azure VMware Solution in the nine-part pattern as dedicated bare-metal VMware private clouds on Azure, for lift-and-shift of vSphere estates, with host-based pricing, minimum cluster sizes and the VMware licensing questions that come with it.
18. Level 18: Azure Spring Apps and Its Retirement
    Learners cover Azure Spring Apps in the nine-part pattern with its announced retirement put first, and plan how to move Spring Boot workloads to Azure Container Apps, AKS or App Service.
19. Level 19: Sizing Machines for Compute-Heavy and HPC Work
    Learners size VMs and clusters for compute-intensive and HPC work from measured needs, comparing cores, clock speed, memory bandwidth, memory per core, accelerator memory and interconnect across the families covered so far.
20. Level 20: Pricing Models: On-Demand, Commitments and Spot
    Learners compare pay-as-you-go, reservations, savings plans, Azure Hybrid Benefit and Spot pricing on dated example prices, and learn which levers apply to which compute services.
21. Level 21: Worked Total Cost: One Workload, Every Price Model
    Build a dated, line-by-line monthly and three-year total cost for one compute-heavy workload. It covers compute, storage, networking, licensing and operations effort, and compares on-demand, commitment, spot and mixed pricing.
22. Level 22: Fair Benchmarking Across Services and Clouds
    Plan and run benchmarks that give a fair comparison of compute options on Azure and other clouds. Control for size, region, placement, storage, network and run-to-run variation, and report cost per unit of useful work.
23. Level 23: Networking That Compute Depends On
    Learn the network decisions every compute choice relies on: virtual networks and subnet sizing, private endpoints, egress and NAT, load balancing, Accelerated Networking, and low-latency InfiniBand placement for tightly coupled HPC.
24. Level 24: Storage That Compute Depends On
    Match storage to compute: managed disk tiers, ephemeral OS disks, Blob, Azure Files, Azure NetApp Files and Azure Managed Lustre. Learn their throughput, IOPS and cost trade-offs for data-heavy and HPC workloads.
25. Level 25: Identity, Access and Secrets for Compute
    Secure every compute service with Microsoft Entra ID, managed identities, Azure RBAC and Key Vault. Remove stored credentials and set least-privilege access for pipelines, schedulers and workloads.
26. Level 26: Migration Paths onto Azure Compute
    Choose and plan a migration path for an existing workload, whether rehost, replatform, refactor or replace. Use Azure Migrate assessments, and plan the move from retiring or ill-fitting services to their nearest alternatives.
27. Level 27: The Weighted Selection Scorecard
    Turn the standard workload profile into a weighted scorecard, using dimensions that can be scored on any cloud. Use it to rank compute options consistently, run sensitivity checks and record why you made the choice.
28. Level 28: Scoring Azure Against Another Cloud
    Apply the scorecard to Azure and a second provider (AWS or Google Cloud) for the same workload, using equivalent services, dated price and quota evidence and the same benchmarks, so the comparison is fair rather than vendor-led.
29. Level 29: Final Challenge: Compute Decision Workshop Exam
    Run a full decision workshop on a compute-intensive workload. Profile it, shortlist and benchmark options, cost them, score them across clouds and deploy the chosen service. Finish with a one-page selection record.

Access

The first 2 levels are free with a free account. Every level, the podcast edition and the AI tutor come with All Access at £4.99/month or any Creator plan — see pricing.

Start free →