Cloud Virtualization Technology – Hypervisors, VMs, Containers, and Management

Introduction to Cloud Virtualization

Virtualization is the abstraction of physical computing resources—CPU, memory, storage, and network—into logical units that can be independently managed and isolated. In cloud computing, virtualization is the foundation that allows multi‑tenant infrastructure, rapid provisioning, and elastic scaling.

Key Concepts

• Virtual Machine (VM): A software emulation of a complete computer system that runs a guest operating system (OS) on top of a hypervisor.
• Hypervisor (Virtual Machine Monitor, VMM): Software that creates, runs, and manages VMs by translating guest OS requests into host hardware operations.
• Container: A lightweight, OS‑level virtualization that packages an application and its dependencies into a single executable unit, sharing the host kernel.
• Orchestration: Automated management of container lifecycles, scaling, networking, and fault tolerance, typically via platforms like Kubernetes.

Hypervisor Types

flowchart TD
    A["Physical Server"] --> B["Type‑1 Hypervisor"]
    B --> C["VM1"]
    B --> D["VM2"]
    B --> E["VM3"]
    A --> F["Type‑2 Hypervisor"]
    F --> G["VM4"]
    F --> H["VM5"]

How a Type‑1 Hypervisor Works

1. Hardware Abstraction Layer (HAL): The hypervisor directly interfaces with CPU, memory, I/O devices.
2. Virtual Machine Monitor (VMM): Intercepts privileged instructions from guest OS, translating them into host operations.
3. Resource Scheduler: Allocates CPU cycles, memory pages, and I/O bandwidth to each VM.
4. Device Emulation: Provides virtual devices (disk, NIC, GPU) to guests.

Virtual Machine Lifecycle – Worked Example

Consider a user launching a Windows Server VM on VMware ESXi.

Figure 2: VM lifecycle stages in a hypervisor.

Virtual Machine Architecture

A VM consists of:

• Virtual CPU (vCPU): Logical CPU cores mapped to physical cores.
• Virtual Memory: Guest OS sees a contiguous memory space; hypervisor maps pages to host RAM.
• Virtual Disk: File or block device that represents the VM’s storage.
• Virtual NIC: Emulated network interface connected to a virtual switch.

classDiagram
    class Hypervisor {
        +allocateResources()
        +scheduleCPU()
        +translateIO()
    }
    class VM {
        +vCPU
        +vMemory
        +vDisk
        +vNIC
    }
    Hypervisor --> VM : manages

Storage Virtualization

• Virtual Disk Formats: VMDK (VMware), QCOW2 (QEMU/KVM), VHDX (Hyper‑V).
• Thin Provisioning: Disk space allocated on demand, saving physical storage.
• Snapshots & Clones: Point‑in‑time copies for backup or rapid deployment.

Network Virtualization

• Virtual Switch (vSwitch): Connects VMs to each other and to the physical network.
• Virtual NIC (vNIC): Each VM has one or more vNICs, each with a MAC address.
• Overlay Networks: VXLAN, NVGRE encapsulate VM traffic over physical infrastructure.

Containerization – Docker & Images

Containers differ from VMs in that they share the host kernel and use OS‑level isolation (cgroups, namespaces).

Docker Image Layers

A Docker image is built from a series of layers, each representing a filesystem change.

sequenceDiagram
    participant User
    participant Docker
    User->>Docker: docker build
    Docker->>Docker: Layer 1 (FROM base)
    Docker->>Docker: Layer 2 (RUN apt‑get install)
    Docker->>Docker: Layer 3 (COPY app)
    Docker->>Docker: Image complete

Container vs VM Comparison

Orchestration – Kubernetes Architecture

graph TD
    A["Client"] --> B["API Server"]
    B --> C["Controller Manager"]
    B --> D["Scheduler"]
    B --> E["etcd"]
    C --> F["Replication Controller"]
    D --> G["Node"]
    G --> H["Container Runtime"]
    G --> I["Kubelet"]
    G --> J["Kube Proxy"]
    subgraph Cluster
        G
    end

Kubernetes manages container workloads across a cluster of nodes, providing scaling, self‑healing, and service discovery.

Advantages & Disadvantages

Applications in Cloud and Edge

• Public Clouds: AWS EC2 (VMs), Google Compute Engine (VMs), Azure VMs.
• Private Clouds: OpenStack (VMs + containers), VMware vSphere.
• Edge Computing: Virtualized network functions (VNFs) on 5G base stations.
• DevOps: Docker for CI pipelines, Kubernetes for production deployments.

In the real world

1. eSewa (Digital Wallet) – Uses Docker containers to host microservices (payment processing, user authentication). Each service runs in its own container, sharing the host kernel, enabling rapid scaling during peak transaction periods.
2. Pathao (Logistics & Ride‑Sharing) – Deploys a Kubernetes cluster on Google Cloud to manage its delivery‑tracking microservices. Autoscaling pods handle fluctuating demand during festivals.
3. Ncell (Telecom) – Virtualizes network functions (e.g., BSS/OSS) using VMware ESXi. Virtual routers and firewalls run as VMs, allowing rapid deployment of new services without physical hardware changes.

Worked Example – Daraz Order Queue

Daraz’s order processing system uses a Kubernetes deployment with 3 replicas of the order‑processor pod. Each pod consumes messages from a RabbitMQ queue. When the queue length exceeds 200 messages, the Horizontal Pod Autoscaler (HPA) scales replicas to 6, reducing average processing time from 5 s to 1 s. The system’s resilience is ensured by Kubernetes’ self‑healing: if a pod crashes, it is automatically restarted.

Exam tip

• Understand the difference between Type‑1 and Type‑2 hypervisors – be able to list advantages and typical use cases.
• Explain the VM lifecycle – especially snapshotting and live migration.
• Compare VMs and containers – focus on isolation, performance, and use‑case suitability.
• Describe Kubernetes architecture – know the role of API server, scheduler, controller manager, etc.
• Be ready to draw a simple hypervisor or Kubernetes diagram – use mermaid or figure blocks as shown.

Typical enterprise server rack used in data centers (Image: Abigor, CC BY-SA 3.0, via Wikimedia Commons) Illustration of a virtual machine running inside a hypervisor (Image: Zippie666 at Dutch Wikibooks, GPL, via Wikimedia Commons)