Microsoft Azure: Cloud Architecture Case Studies

Define a practical architectural roadmap to deliver fast, secure, reliable, and maintainable systems by understanding system and non-functional requirements, mapping components, selecting the technology stack, and documenting the architecture.

Define system's requirements by understanding goals, high level tasks such as view telemetry data, workflows, services, and interface elements, then meet the system analyst to discuss needs and schedule sessions.

Architects define non-functional requirements, shaping system attributes like concurrency, load, data volumes, and performance. Understand these requirements early, as they influence many architectural decisions in cloud architecture case studies.

Map the components of the system to understand its parts and communicate your scope to clients. Keep the component map non-technical and independent of platform, tools, and database choices.

Collaborate with the development team to select the platform and technologies for the system, including back end, front end, and data store, with rational evaluation to support microservices.

Design the architecture that glues requirements, non-functional requirements, components, and technology stack into a fast, secure, reliable, and maintainable system, using loose coupling, stateless scaling, caching and messaging as blocks.

Describe the architecture document that captures the full development process and provides a complete view for developers and management, while outlining its structure, content, and ways to maximize its value.

Support the team beyond document delivery to prevent the architecture from becoming a glorified paperweight. Stay with developers as the architecture changes and you navigate dilemmas and talks.

Architects collaborate with clients, system analysts, and developers to define non-functional requirements and build the architecture together, preventing future disagreements and earning buy-in for a practical solution.

Understand system and non-functional requirements, map components, and select the technology stack to design architecture for each case study, while excluding architectural documents and team support.

The lecture explains how on-site servers create underutilized compute and outages on peak days like Black Friday; cloud computing offers resources to solve this problem.

Understand how the cloud provides compute, storage, and networking via providers operating global data centers with hundreds of services, including AI, IoT, and Kubernetes.

Explore the five characteristics of cloud computing—on-demand self-service, broad network access, resource pooling, rapid elasticity, and measured service—and how automated provisioning, global access, and pay-for-use optimize resources.

Explore IaaS, PaaS, and SaaS by comparing who controls the infrastructure, platform, or software; see how virtual machines, web apps, and Office 365 illustrate each model.

Explore public, private, and hybrid clouds; public clouds are internet-based and managed by major providers, while private clouds stay on premises and hybrids connect them to the public cloud.

Compare cloud providers and their public services (iaas, paas, saas) from amazon web services, microsoft azure, and google cloud platform, noting azure’s rapid growth and its relevance to cloud architecture.

Explore Microsoft Azure, announced in 2008 and released in 2010, now the second largest public cloud, offering the largest variety of services after starting with PaaS and later adding IaaS.

Explore how Azure regions map the global data center footprint, covering about sixty regions with availability zones and paired regions for resilient geo-redundancy.

Explore how Azure provides hundreds of cloud services across regions and zones, covering virtual machines, databases, networks, AI algorithms using central user management.

Explore how cloud architecture differs from classic design by emphasizing existing services and cost, and learn how to implement these differences in practice.

Use existing cloud services to simplify deployment, security, and scaling. Show how Azure SQL and Azure Functions offer managed alternatives to VM-based setups, with caveats about control vs effort.

Explore CapEx vs OpEx in cloud design, showing how pay-as-you-go models like Azure function and storage options trade cost against performance and features.

Explore real world cloud architecture case studies by introducing the company and application, defining functional and non-functional requirements, and mapping the architecture components with a suitable technology stack.

Explore functional and non-functional requirements for a web-based HR system handling salaries, vacations, CRUD, and a legacy external payment interface, including data volume estimates and concurrency.

Map components by assigning a single service to each entity—employees, salary, vacation—and a view service, use a single data store, logging via a queue, and payment via a file interface.

Design a centralized logging service using Azure Monitor and Azure Log Analytics to ingest, query with the Kusto query language, and visualize logs from apps and resources.

Explore the view service that delivers static files (HTML, CSS, JavaScript) with no business logic, as a static web app in Azure, and compare app service with static web apps.

Design and implement the employees service as a web api using dotnet core on Azure App Service, enabling CRUD operations on employees and documents, with Azure SQL and storage.

Design a salary service on Azure web api using dotnet core app service, exposing endpoints to add, remove, view, and approve salary change requests; leverage auto scaling for reliability.

Explore the vacation service web API built on dotnet core and azure app service, enabling HR to set available days and employees to reduce days within a three-tier architecture.

Implement a batch payment interface that retrieves salary data and sends it to a payment system via an app service web job, evaluating Azure batch options and monitoring needs.

Secure the Dundalli cloud by enforcing encryption and TLS for Azure storage and Azure SQL, then implement an application gateway with WAF and strict access restrictions.

Present a complete architecture diagram for the Dhandoli application and review the total cloud cost of 848 dollars per month, with a downloadable Azure Calculator cost estimation in Excel.

Introduce Mascot Auto’s autonomous vehicle system and the need for a reliable telemetry reception and data display solution to improve car performance and safety.

Identify functional and non-functional requirements for a web-based telemetry system, plan one-week operational data retention, and ensure near real-time dashboards and bi analytics.

Map the telemetry architecture from cars through a thin gateway and telemetry pipeline to the processor and viewer, with extract-transform-load moving data to the data warehouse and archive.

Design a nodejs-based telemetry gateway that receives TCP telemetry from cars and pushes data to the pipeline. Implement it on Linux VMs behind scale set and load balancer in Azure.

The telemetry pipeline queues gateway messages for processing and uses Azure event hubs for high-volume streaming, with a one megabyte max size and seven throughput units.

Design a telemetry viewer web app and web API to query latest telemetry and errors for cars in real-time, using NodeJS backend and Reactjs frontend on Azure app services.

Analyze telemetry data with the business intelligence component to extract insights and display reports, trends, and forecasts such as car failures last month and total distance driven using Power BI.

Secure cloud architectures by blocking database access from unauthorized ip addresses and evaluating firewall or waf deployment in front of app services to balance security and cost.

Present architectural diagram of the massacre cloud system, detailing load balancer, vm scale set, event hubs, cosmos db, storage, app service, and a BI-designed data warehouse; monthly cost around $1,835.

Introduce the collection side of Grow Skull's grocery service, where employees use tablets to mark items as collected or unavailable while customers create online shopping lists.

Identify the functional and non-functional requirements for a web-based tablet workflow that collects lists and transfers them to a payment engine, with offline support and queue-driven intake.

Map the shopping list system components, enabling offline tablets to display and update lists, mark items, and export data to the payment engine via a decoupled list service and queue.

Design a list receiver that ingests shopping lists from a queue and stores them in Azure MySQL via a Java function app, using partitioning and the consumption tier.

Design lists service as web app and api in Azure App Service using Java, with three-layer architecture to query lists, mark items as executed or unavailable, and export payment data.

Design a web-based front end that displays the shopping list, marks items as unavailable or collected, and sends the list to payment, using React Native for cross-device offline support.

Utilize the existing company queue to implement the list data service, delivering shopping list data to the payment system while avoiding new queue deployments.

Assess security risks of publicly accessible databases and unprotected app services, and discuss blocking unauthorized IPs and omitting a WAF on a small read-only app service to save costs.

Conclude the case study with the final architecture diagram using Function App, App Service, and Azure MySQL, then review the total estimated cost and download the detailed cost estimate.

Explore the payroll payment processing case study, introducing a fictitious payroll system that automatically validates, processes, and delivers bank instructions from sources to banks, ensuring reliability and speed.

Understand how to receive, validate, and process diverse files, ensure no data loss, and generate a unified bank payment file under 1-minute latency with seven years of logs.

Map the payroll processing architecture. Use a queue-driven pipeline to ingest files, format to a unified internal format, compute, export to the bank folder, and log seven years of activity.

Design a file handler as an Azure function app that pulls payment files from folders and enqueues them, using .NET Core for performance and ease within a consumption-tier cost model.

Three queues deliver files, the formatted files, and calculated files via event hubs, chosen for durability and load balancing when 1 MB messages exceed storage queue limits.

Validate and format incoming topic files into a unified format with the file formatter. Implement as an Azure function app, enqueue results, and support new formats as they are developed.

The file calculation service consumes queued files, computes data, and outputs new files to a queue, mirroring the file formatter and built with function up under free function ops.

The file exporter service processes files from the queue, stores them in the bank's folder, and finalizes file handling using an Azure function app, illustrating reusable Azure patterns.

Analyze the logs service using Azure Log Analytics for visualizations and Power BI, noting a two-year retention and a $46 monthly cost, with archiving options for seven-year needs.

Review the complete architectural diagram of a payment system using function apps, event hubs, and log analytics on Azure, with a total cost of about $68 per month.

Explore a case study for a security camera system in the cloud, detailing privacy, extreme separation between the clients' tenants, component mapping, cost estimation, and video analysis.

Receive a daily stream of one megabyte photos and analyze for suspicious objects. Notify clients, display security event dashboards and central dashboard, and enforce per-client data separation with authorized access.

Map the system components, including the file receiver, file storage, file analyzer, analysis results database, notifier, and client dashboard, with strict access control and client-only data.

Design the file receiver with a storage account rather than IoT Hub due to lower costs. Cameras upload via SAS token and REST API, storing files in the storage account.

Explore the file analyzer that ingests photos from storage, uses cognitive services' Analyze Image API to detect objects, and stores results in per-client Azure SQL database via a function app.

The notifier uses Azure notification hub to send cross-platform push notifications to Android and iPhone devices. The basic tier covers 10 million pushes for 200,000 devices at $10/month.

Design the client dashboard on Azure to display security event data and enable basic querying, implemented with App Service for a fully managed web app.

Implement access control by authenticating users for the client's tenant with Azure AD, a fully managed identity provider integrated with app service, using the free tier up to 500,000 objects.

Build a secure management dashboard for private client data using a fully managed Azure app service and Azure SQL, deployed in a separate subscription for USD 260 per month.

Enforce default database encryption and configure Azure AD to secure data; assess WAF needs, delaying for the client dashboard but preparing for the management dashboard due to sensitive payment data.

Explore the Sky case study's complete architecture with two separate subscriptions for client and company, highlighting SaaS data isolation and monthly costs of about $930 and $333.

Review Azure cloud architecture case studies, including a classic web app, IoT telemetry, tablet data collection, async file processing, and a multi-tenant ai system.