Server, Storage & Network Virtualization: Concepts, Techniques & Real-World Use

Core Concepts: What Is Virtualization?

Virtualization is the process of creating a virtual (rather than physical) version of computing resources—servers, storage, or networks—so that multiple workloads can run on a single physical machine efficiently. In cloud computing, this enables:

• Resource pooling: Sharing hardware among multiple users (tenants).
• Isolation: Each virtual resource operates independently.
• Scalability: Dynamically allocating resources based on demand.

Why Virtualize?

1. Server Virtualization: Turning One Machine into Many

Server virtualization divides a physical server into multiple virtual machines (VMs), each running its own OS and applications.

How It Works

1. Hypervisor (VMM): Software that allocates hardware resources (CPU, RAM, storage) to VMs.
   • Type 1 (Bare-metal): Runs directly on hardware (e.g., VMware ESXi, Microsoft Hyper-V).
   • Type 2 (Hosted): Runs on a host OS (e.g., Oracle VirtualBox, VMware Workstation).
2. VM Components:
   • Guest OS: Runs inside the VM (e.g., Windows Server, Linux).
   • Virtual Hardware: Emulated CPU, RAM, disk, and network interfaces.
   • Hypervisor Layer: Manages resource allocation and isolation.

Real-World Example: eSewa’s Cloud Infrastructure

eSewa, Nepal’s leading digital payment platform, uses server virtualization to:

• Host multiple services (bill payments, remittance, e-commerce) on fewer physical servers.
• Scale VMs during peak hours (e.g., Dashain/Tihar festivals).
• Isolate payment processing VMs from customer-facing VMs for security.

Worked Example: VM Resource Allocation Suppose a cloud provider assigns resources to VMs for a Daraz order processing system:

• VM1 (Web Server): 2 vCPUs, 4GB RAM, 50GB SSD.
• VM2 (Database): 4 vCPUs, 8GB RAM, 200GB HDD.
• VM3 (API Gateway): 1 vCPU, 2GB RAM, 20GB SSD. Question: If the hypervisor allocates 16 vCPUs and 32GB RAM, how much is left for the host OS? Answer: Total allocated = (2+4+1) vCPUs + (4+8+2) GB RAM = 7 vCPUs, 14GB RAM. Remaining = 16–7 = 9 vCPUs, 32–14 = 18GB RAM for the host.

2. Storage Virtualization: Pooling and Abstracting Storage

Storage virtualization combines multiple physical storage devices (HDDs, SSDs, NAS) into a single logical pool, managed by software.

Types of Storage Virtualization

How It Works

1. Storage Area Network (SAN): High-speed network connecting servers to storage arrays.

2. Network-Attached Storage (NAS): Dedicated file storage device on a network.

Real-World Example: Ncell’s Cloud Storage

Ncell uses object storage to:

• Store millions of customer photos/videos in a scalable, cost-effective way.
• Enable auto-scaling during data backups (e.g., nightly uploads).
• Provide disaster recovery via geographically distributed storage.

Worked Example: Storage Allocation for a Bank A bank needs to store 1TB of transaction data. The cloud provider offers:

• Option 1: 1TB HDD ($50/month, 100MB/s speed).
• Option 2: 1TB SSD ($150/month, 500MB/s speed). Question: Which should the bank choose for high-frequency transactions? Answer: Option 2 (SSD) because:
• Faster read/write speeds reduce latency.
• Critical for real-time processing (e.g., loan approvals, fund transfers).

3. Network Virtualization: Software-Defined Networks (SDN)

Network virtualization decouples the network’s physical hardware from its control logic, allowing dynamic configuration via software.

Key Techniques

How SDN Works

1. Control Plane: Software (SDN controller) decides where traffic goes.
2. Data Plane: Physical switches/routers forward traffic based on controller rules.
3. Northbound API: Applications request network services (e.g., "Create a new VLAN").
4. Southbound API: Controller communicates with switches (e.g., OpenFlow).

sequenceDiagram
    participant App as Application
    participant Controller as SDN Controller
    participant Switch as Physical Switch
    App->>Controller: Request new VLAN
    Controller->>Switch: Install flow rules (OpenFlow)
    Switch-->>App: Forward traffic via new VLAN

Real-World Example: Pathao’s Dynamic Routing

Pathao uses network virtualization to:

• Isolate rider and driver apps in separate VLANs for security.
• Dynamically reroute traffic during peak hours (e.g., evening rush) via SDN.
• Optimize latency for real-time GPS updates.

Worked Example: VLAN Configuration A university has 3 departments (CS, IT, EE) sharing a network. Assign VLAN IDs:

• CS: VLAN 10 (192.168.10.0/24)
• IT: VLAN 20 (192.168.20.0/24)
• EE: VLAN 30 (192.168.30.0/24) Question: If a CS student (IP: 192.168.10.50) tries to access an IT server (192.168.20.100), what happens? Answer: The router drops the packet because:
• VLANs are isolated by default.
• Inter-VLAN routing requires explicit configuration (e.g., router ACLs or layer-3 switch).

4. Integration in Cloud Environments

Cloud providers combine all three virtualizations to deliver Infrastructure as a Service (IaaS).

Cloud Virtualization Stack

Real-World Example: Google Cloud’s Global Network

Google uses:

• Server Virtualization: Millions of VMs running on Borg (Google’s custom hypervisor).
• Storage Virtualization: Colossus (object storage) for user data (Gmail, Drive).
• Network Virtualization: B4 (SDN-based global network) for low-latency routing.

Worked Example: Cost Calculation for a Cloud VM A startup deploys a VM with:

• 2 vCPUs, 4GB RAM, 100GB SSD.
• Google Cloud pricing: $0.05/hour for vCPU, $0.01/GB RAM, $0.10/GB SSD. Question: Monthly cost (730 hours)? Answer:
• vCPU: 2 × $0.05 × 730 = $73
• RAM: 4 × $0.01 × 730 = $29.20
• SSD: 100 × $0.10 = $10 Total = $73 + $29.20 + $10 = $112.20/month

In the Real World

1. Khalti’s Payment Processing

   • Uses server virtualization to isolate transaction VMs from customer-facing VMs.
   • Storage virtualization pools SSDs for high-speed fraud detection.

2. Daraz’s Order Fulfillment

   • Network virtualization (SDN) dynamically routes traffic during sales (e.g., Dashain).
   • Container orchestration (Kubernetes) manages microservices, but underlying VMs use server virtualization.

3. NTC’s Network Monitoring

   • SDN controllers analyze traffic patterns to detect DDoS attacks.
   • VLANs segment internal and customer networks.

Exam Tip

1. Definitions: Know the difference between Type 1 vs. Type 2 hypervisors, SAN vs. NAS, and VLAN vs. VPN.
2. Diagrams: Be ready to draw:
   • Hypervisor-VM relationships.
   • SAN/NAS architectures.
   • SDN control/data planes.
3. Calculations: Practice:
   • VM resource allocation.
   • Storage cost comparisons (HDD vs. SSD).
   • VLAN/IP subnetting.
4. Applications: Link concepts to Nepali companies (eSewa, Ncell, Daraz) or global platforms (AWS, Google Cloud).
5. Shortcomings: Discuss limitations like:
   • Virtualization overhead (hypervisor latency).
   • Storage bottlenecks in SANs.
   • SDN complexity in legacy networks.