Cloud Computing Cloud Computing Foundations Syllabus

What does 'serverless' mean, and does it mean there are no servers?

Serverless means the developer does not provision or manage servers—the cloud provider handles infrastructure, scaling, and capacity automatically. Servers still exist; they are just abstracted away and billed per use (e.g., AWS Lambda, Azure Functions).

Name two defining traits of FaaS billing and scaling.

(1) Billing is granular and usage-based—you pay only for execution time/invocations, not idle capacity; (2) automatic scaling including scale-to-zero when there are no requests.

What is the shared responsibility model in cloud computing?

A security framework dividing duties between provider and customer: the provider is responsible for security 'of' the cloud (physical infrastructure, hardware, hypervisor), while the customer is responsible for security 'in' the cloud (data, access management, and configurations, to varying degrees by service model).

How does customer responsibility change across IaaS, PaaS, and SaaS in the shared responsibility model?

Customer responsibility decreases from IaaS to SaaS. In IaaS the customer manages OS, runtime, apps, and data; in PaaS only apps and data; in SaaS mainly data, user access, and configuration. The provider assumes correspondingly more responsibility toward SaaS.

In the shared responsibility model, who is always responsible for data classification and user access management regardless of service model?

The customer. Across IaaS, PaaS, and SaaS, the customer always retains responsibility for their data, endpoints/accounts, and identity and access management.

What are the four main cloud deployment models?

Public cloud, Private cloud, Hybrid cloud, and Community cloud.

Define public cloud and state its ownership model.

A public cloud provides computing resources over the internet, owned and operated by a third-party provider and shared among multiple tenants/organizations (multi-tenant). Examples: AWS, Microsoft Azure, Google Cloud Platform.

Define private cloud and give its key advantage.

A private cloud is cloud infrastructure dedicated to a single organization, hosted on-premises or by a third party. Its key advantage is greater control, customization, security, and compliance, at higher cost.

Define hybrid cloud and its primary benefit.

A hybrid cloud combines private and public clouds (and/or on-premises) with orchestration allowing data and applications to move between them. Benefit: flexibility—keep sensitive workloads private while using public cloud for scalable/burst needs.

What is 'cloud bursting' in a hybrid cloud?

Cloud bursting is a hybrid deployment technique where an application runs in a private cloud/data center and 'bursts' into a public cloud for extra capacity when demand spikes beyond local resources.

What is a multi-cloud strategy, and how does it differ from hybrid cloud?

Multi-cloud uses services from two or more public cloud providers (e.g., AWS + Azure). Unlike hybrid cloud, it does not necessarily combine public and private; its focus is using multiple providers to avoid lock-in, optimize cost, and improve resilience.

Give two main reasons organizations adopt a multi-cloud strategy.

(1) To avoid vendor lock-in and gain negotiating leverage; (2) to increase resilience/availability and pick best-of-breed services or optimize cost/performance across providers.

Define community cloud and who uses it.

A community cloud is shared infrastructure provisioned for exclusive use by a specific community of organizations with common concerns (e.g., mission, security, compliance requirements). Example: a cloud shared among government agencies or healthcare providers.

What is a hypervisor, and what is its core function?

A hypervisor (virtual machine monitor, VMM) is software/firmware that creates and runs virtual machines by abstracting and allocating physical hardware resources, allowing multiple VMs (each with its own OS) to run on one physical host.

Distinguish Type 1 and Type 2 hypervisors.

Type 1 (bare-metal) runs directly on physical hardware with no host OS (e.g., VMware ESXi, Microsoft Hyper-V, Xen, KVM)—higher performance, used in data centers. Type 2 (hosted) runs as an application on top of a host OS (e.g., VirtualBox, VMware Workstation)—used on desktops.

Which hypervisor type generally offers better performance and why?

Type 1 (bare-metal), because it runs directly on the hardware without an intervening host operating system, reducing overhead and giving VMs more direct access to physical resources.

What is a virtual machine (VM)?

A VM is a software-based emulation of a physical computer that runs its own complete guest operating system and applications, using virtualized hardware resources allocated by a hypervisor, while sharing the underlying physical host with other VMs.

What is a VM image (machine image)?

A VM image is a static template/snapshot file containing a preconfigured operating system, software, and settings, used to create/instantiate new virtual machines quickly and consistently (e.g., an AWS AMI).

What is a container in cloud computing?

A container is a lightweight, portable, executable package that bundles an application with its dependencies and libraries, running in an isolated user space while sharing the host operating system's kernel via OS-level virtualization.

State the fundamental architectural difference between containers and virtual machines.

VMs virtualize hardware—each includes a full guest OS running on a hypervisor. Containers virtualize the OS—they share the host's kernel and package only the app and its dependencies, making them far lighter.

Compare containers and VMs on size, startup time, and isolation.

Containers: megabytes in size, start in seconds/milliseconds, process-level isolation (weaker). VMs: gigabytes, start in minutes, strong hardware-level isolation with a full separate OS.

Why are containers more portable and resource-efficient than VMs?

Because they share the host OS kernel and don't bundle a full guest OS, containers have much smaller footprints, lower overhead, and can be moved consistently across environments; more of them can run on the same hardware than VMs.

Distinguish hardware-level virtualization from OS-level virtualization.

Hardware-level virtualization uses a hypervisor to emulate physical hardware so multiple full guest OSes (VMs) can run on one host. OS-level virtualization uses the single host kernel to create isolated user-space instances (containers) that share that kernel—no separate guest OS.

Which virtualization type do containers use, and what do they share as a result?

Containers use OS-level virtualization; as a result they share the host operating system's kernel, avoiding the overhead of running separate guest operating systems.

What technology is commonly used to orchestrate containers at scale, and what does it do?

Kubernetes. It automates deployment, scaling, load balancing, self-healing, and management of containerized applications across a cluster of machines.

In cloud pricing, express a simple pay-as-you-go monthly compute cost as a formula.

$$C = r \times h \times n$$ where $C$ is total cost, $r$ is the hourly rate per instance, $h$ is hours of usage, and $n$ is the number of instances.

Define scalability and distinguish vertical from horizontal scaling.

Scalability is a system's ability to handle increased load. Vertical scaling (scale up) adds more power (CPU/RAM) to an existing machine; horizontal scaling (scale out) adds more machines/instances to distribute the load.

What is multi-tenancy, and which deployment model most relies on it?

Multi-tenancy is an architecture where a single instance of software/infrastructure serves multiple customers (tenants) with logically isolated data. The public cloud most relies on it, since resources are pooled and shared across many organizations.

How does availability, often expressed as a percentage in SLAs, relate to allowed downtime?

Availability is the percentage of time a service is operational, e.g. an SLA of $99.9\%$ ('three nines') permits about $8.76$ hours of downtime per year, while $99.99\%$ ('four nines') permits about $52.6$ minutes per year: $$\text{Downtime} = (1 - A) \times \text{total time}.$$