Cloud computing has fundamentally changed how organizations provision, manage, and scale their IT infrastructure. Rather than maintaining physical hardware on-premises, teams can now access compute, storage, networking, and application services on demand over the internet—paying only for what they use. Understanding the cloud computing stack and the vendors that dominate each layer is essential for any DevOps or platform engineering team making infrastructure decisions.
What Is the Cloud Computing Stack?
The cloud computing stack refers to the layered set of services that together form a complete cloud offering. Each layer builds on the one below it, abstracting away progressively more infrastructure complexity. The three canonical layers are:
- Infrastructure as a Service (IaaS) – Raw compute, networking, and storage resources delivered over the internet. Teams retain full control over the operating system and above.
- Platform as a Service (PaaS) – A managed environment for developing, testing, and deploying applications without managing the underlying infrastructure. The provider handles the OS, runtime, middleware, and scaling.
- Software as a Service (SaaS) – Fully managed applications delivered via a browser or API. The provider is responsible for everything from infrastructure to the application layer itself.
These three layers are often visualized as a stack because each higher tier consumes services from the tier below it. A SaaS product, for example, typically runs on PaaS or IaaS underpinnings provided by a hyperscaler.
Layer 1: Infrastructure as a Service (IaaS)
IaaS gives engineers the most control over their environment. Typical IaaS resources include virtual machines, block and object storage, virtual networking, load balancers, and bare-metal servers. The consumer manages everything from the guest operating system upward—patching, runtime configuration, security hardening, and application deployment all remain the team’s responsibility.
Common use cases:
- Lift-and-shift migration of legacy applications
- High-performance computing and batch workloads
- Custom network topologies and security configurations
- Disaster recovery and backup infrastructure
Key IaaS providers: Amazon Web Services (EC2, S3, VPC), Microsoft Azure (Virtual Machines, Blob Storage, VNet), Google Cloud Platform (Compute Engine, Cloud Storage), IBM Cloud, Rackspace, and OVHcloud.
Layer 2: Platform as a Service (PaaS)
PaaS removes the burden of OS and middleware management, letting development teams focus on writing and deploying code. The platform handles auto-scaling, load balancing, and often integrates CI/CD pipelines, managed databases, and monitoring out of the box.
Common use cases:
- Web and mobile application development and hosting
- Microservices deployment on managed Kubernetes clusters
- Serverless function execution
- Managed database and messaging services
Key PaaS providers: AWS Elastic Beanstalk and Lambda, Azure App Service and Azure Functions, Google App Engine and Cloud Run, Heroku (Salesforce), IBM Cloud Foundry, and SAP Business Technology Platform.
Layer 3: Software as a Service (SaaS)
SaaS is the most abstracted layer. End users interact with a fully managed application—there is no infrastructure or runtime to configure. From a DevOps perspective, SaaS is relevant both as a consumer (using tools like GitHub, Datadog, or PagerDuty) and as a target delivery model when building products for customers.
Common use cases:
- CRM and ERP systems
- Collaboration and productivity tools
- Monitoring, observability, and ITSM platforms
- Developer tooling (source control, CI/CD, issue tracking)
Key SaaS providers: Salesforce (CRM), Microsoft 365, Google Workspace, SAP, Oracle Cloud Applications, and Avaya cloud communications.
Major Cloud Vendors Across the Stack
Most hyperscalers operate across all three layers simultaneously. Below is a concise overview of the leading vendors and their primary strengths:
| Vendor | Primary Layer(s) | Notable Strengths |
|---|---|---|
| Amazon Web Services (AWS) | IaaS, PaaS, SaaS | Broadest service catalogue, global region coverage, mature tooling ecosystem |
| Microsoft Azure | IaaS, PaaS, SaaS | Deep enterprise and Active Directory integration, hybrid cloud with Azure Arc |
| Google Cloud Platform (GCP) | IaaS, PaaS, SaaS | Kubernetes origins (GKE), BigQuery analytics, AI/ML tooling |
| IBM Cloud | IaaS, PaaS, SaaS | Enterprise mainframe integration, Red Hat OpenShift, regulated industries |
| Salesforce | PaaS, SaaS | CRM leadership, Force.com / Heroku developer platform |
| SAP | PaaS, SaaS | ERP and business applications, SAP BTP integration |
| Oracle Cloud | IaaS, PaaS, SaaS | Database workloads, Oracle application suite |
| Rackspace | IaaS, Managed Services | Managed cloud and colocation, multi-cloud managed services |
| OVHcloud | IaaS | European data sovereignty, competitive bare-metal pricing |
| Avaya | SaaS | Cloud-based communications and contact centre solutions |
Choosing the Right Layer for Your Workload
Selecting the appropriate cloud layer is a trade-off between control and operational overhead:
- Choose IaaS when you need maximum control over the OS, network configuration, or custom hardware (e.g., GPU instances, specific compliance requirements).
- Choose PaaS when developer velocity matters more than infrastructure control—managed runtimes, auto-scaling, and built-in CI/CD integrations reduce toil significantly.
- Choose SaaS when the capability is not a core differentiator for your business and a third-party product meets your requirements—let the vendor handle operations entirely.
Many production architectures combine all three layers. A typical e-commerce platform might run compute on IaaS virtual machines or containers (IaaS/PaaS), use a managed database service (PaaS), and integrate a SaaS CRM for customer data—all within the same cloud account.
Market Context and Growth Drivers
Adoption of cloud services has accelerated across every industry vertical, driven by several converging factors:
- Operational agility: Cloud provisioning reduces the time from idea to running infrastructure from weeks to minutes.
- Cost model shift: Moving from capital expenditure (CapEx) to operational expenditure (OpEx) improves financial flexibility.
- Global reach: Hyperscaler regions and edge locations allow teams to deploy closer to end users worldwide.
- Managed services breadth: Vendors continually expand their PaaS and SaaS catalogues, reducing the engineering effort required for common capabilities such as databases, message queues, and ML inference.
- Regulatory and compliance tooling: Cloud providers now offer certified environments for regulated industries (HIPAA, FedRAMP, ISO 27001, PCI DSS), lowering the barrier for enterprise adoption.
Key Takeaways for DevOps Teams
- Understand which layer each service in your architecture sits on—it directly determines your shared responsibility boundary with the vendor.
- Evaluate managed PaaS options before defaulting to IaaS; reduced operational overhead often outweighs the loss of low-level control.
- Multi-cloud strategies can avoid vendor lock-in but introduce their own complexity—ensure your team has the tooling (Terraform, Kubernetes, etc.) to manage it consistently.
- Security and compliance ownership shifts depending on the layer: IaaS gives you the most responsibility; SaaS gives you the least. Map this explicitly in your threat model.