Unit 5 — Contemporary Computing Paradigms, Cloud Systems & Virtualization
1. Evolution of Computing Paradigms
Computing has undergone profound structural shifts over the past seven decades, transitioning from centralized monolithic mainframes to decentralized, globally distributed elastic cloud networks:
[ Centralized Mainframes ] ---> [ Client-Server Networks ] ---> [ Distributed Grid/Clusters ] ---> [ Elastic Cloud & Edge ]
(Single Large Computer) (Dedicated Server + PC) (Shared Computational Grid) (On-Demand Scalable Utility)
1.1 Classification of Contemporary Computing Paradigms
| Computing Paradigm | Architecture & Organization | Key Characteristics | Real-World Use Cases |
|---|---|---|---|
| Centralized Computing | All computational processing, memory, and data storage reside on a single powerful central machine (Mainframe). Terminals are "dumb". | Single point of failure; zero client-side processing; high hardware cost. | Historical enterprise banking, airline reservation legacy mainframes. |
| Client–Server Computing | Workloads are partitioned between service providers (servers) and service requesters (clients) over a network. | Centralized data management; scalable servers; clients handle presentation. | Web browsing (HTTP/HTTPS), email (SMTP/IMAP), database queries (SQL). |
| Peer-to-Peer (P2P) | Every connected node (peer) possesses equivalent privileges and acts as both a supplier and consumer of resources. No central server. | Highly resilient; decentralized; self-scaling bandwidth; immune to single-node failures. | BitTorrent file sharing, Blockchain networks (Bitcoin, Ethereum). |
| Cluster Computing | A collection of interconnected independent, homogeneous computers (nodes) tightly linked over high-speed LANs acting as a single unified computer. | Centralized management; high availability; load balancing; high performance per dollar. | High-Performance Computing (HPC), financial risk modeling, rendering farms. |
| Grid Computing | A federated network of geographically dispersed, loosely coupled, heterogeneous computer systems collaborating on massive scientific computations. | Harnesses idle compute cycles worldwide; decentralized ownership; batch job scheduling. | Large Hadron Collider (CERN), SETI@home, climate forecasting simulation. |
| Cloud Computing | On-demand network delivery of configurable computing resources (servers, storage, databases, networking) via self-service APIs with utility pricing. | Rapid elasticity, broad network access, resource pooling, pay-as-you-go billing. | AWS, Microsoft Azure, Google Cloud Platform (GCP). |
| Edge & Fog Computing | Processing data locally at or near the physical source (sensors, IoT gateways, cell towers) rather than transmitting all raw telemetry to distant cloud centers. | Ultra-low latency ($< 5\text{ms}$), massive bandwidth conservation, offline operation. | Autonomous self-driving vehicles, smart factory robotics, remote medical telemetry. |
2. Cloud Computing Architecture & Decoupled Systems
According to the official NIST Definition (SP 800-145):
"Cloud computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources that can be rapidly provisioned and released with minimal management effort or service provider interaction."
2.1 The 5 Essential Characteristics of Cloud Computing
- On-Demand Self-Service: Customers can provision compute instances and storage automatically through a web console or API without human provider intervention.
- Broad Network Access: Capabilities are available over standard internet networks and accessible through heterogeneous thin/thick client platforms (laptops, phones, workstations).
- Resource Pooling: Provider resources are pooled to serve multiple consumers using a multi-tenant model, dynamically assigning physical and virtual resources according to consumer demand.
- Rapid Elasticity: Resources can be elastically provisioned and released—in some cases automatically (autoscaling)—to scale rapidly outward during demand spikes and inward during quiet hours.
- Measured Service: Resource usage is monitored, controlled, and reported transparently, charging consumers based on exact consumption (utility pay-as-you-go billing).
2.2 Decoupled Architectures: Monoliths vs. Microservices
Modern cloud applications abandon fragile monolithic designs in favor of decoupled architectures:
- Monolithic Architecture: Entire application (UI, business logic, database connectors) is built and deployed as a single unified executable binary. If one component fails or experiences high traffic, the entire application degrades.
- Microservices Architecture: The application is decomposed into small, modular, independent services (e.g., Auth Service, Payment Service, Catalog Service). Each service runs in its own process, manages its own private database, and communicates via lightweight REST APIs or message queues.
- Asynchronous Message Queues: Decouple producers from consumers (e.g., RabbitMQ, Apache Kafka, AWS SQS), buffering requests so bursts in user traffic do not crash downstream databases.
- Serverless Computing (FaaS - Function as a Service): Developers write and deploy standalone function code (e.g., AWS Lambda, Google Cloud Functions) without configuring or managing server operating systems. The cloud provider spins up containers in milliseconds on-demand and bills purely for execution runtime down to the millisecond.
3. Cloud Service Models (IaaS, PaaS, SaaS)
Cloud computing delivers services across three standard functional abstraction layers:
+-------------------------------------------------------------+
| SaaS (Software as a Service) |
| End-user applications (Gmail, Office 365, Salesforce) |
+-------------------------------------------------------------+
| PaaS (Platform as a Service) |
| Runtimes & Databases (Elastic Beanstalk, Heroku, GAE) |
+-------------------------------------------------------------+
| IaaS (Infrastructure as a Service) |
| Raw Compute, VMs & Storage (AWS EC2, Azure VMs, S3) |
+-------------------------------------------------------------+
3.1 Detailed Service Models Comparison
| Dimension | IaaS (Infrastructure as a Service) | PaaS (Platform as a Service) | SaaS (Software as a Service) |
|---|---|---|---|
| What the Provider Supplies | Physical servers, virtualization layer, network switches, data center storage. | Everything in IaaS + Operating System, programming runtimes (Node, Python, Java), database servers. | Complete, fully managed software application ready for immediate end-user operation. |
| What the Customer Manages | OS installation, security patches, network firewalls, runtime installation, code. | Application source code, application data, and environment configuration. | Nothing technical; customer only manages user accounts and organizational data. |
| Target Audience | System Administrators, Cloud Architects, DevOps Engineers. | Software Developers, Web Programmers. | Business end-users, students, consumers. |
| Core Advantage | Maximum control over operating system and hardware configuration. | Rapid development; zero server administration or OS patching overhead. | Accessible instantly from any browser; zero local installation or software maintenance. |
| Real-World Examples | Amazon EC2, Google Compute Engine, Microsoft Azure VMs, DigitalOcean Droplets. | AWS Elastic Beanstalk, Heroku, Google App Engine, Firebase App Hosting. | Google Workspace (Gmail, Drive), Microsoft 365, Salesforce, Zoom, Dropbox. |
3.2 The Shared Responsibility Model
Component IaaS PaaS SaaS
------------------------------------------------------------------
Customer Data Customer Customer Customer
User Access & IAM Customer Customer Customer
Applications Customer Customer Provider
Runtime & Middleware Customer Provider Provider
Operating System & Patches Customer Provider Provider
Virtualization Layer Provider Provider Provider
Physical Hardware & Network Provider Provider Provider
Data Center Facilities Provider Provider Provider
4. Cloud Deployment Models
A deployment model specifies the ownership, accessibility, and physical hosting environment of cloud infrastructure:
- Public Cloud:
- Owned and operated by a third-party commercial cloud provider (AWS, Azure, Google Cloud).
- Infrastructure is shared across multiple tenants over the public internet.
- Pros: Zero capital expenditure (CapEx), infinite scalability, high reliability.
- Cons: Multi-tenant shared infrastructure; potential regulatory compliance concerns.
- Private Cloud:
- Dedicated exclusively to a single organization (enterprise, bank, government agency).
- Can be physically hosted in an on-premises company data center or managed by a third party.
- Pros: Ultimate data privacy, total security control, compliance with sovereign data laws.
- Cons: High upfront capital equipment cost; requires dedicated internal IT maintenance staff.
- Hybrid Cloud:
- Seamlessly integrates a private cloud / on-premises data center with a public cloud via encrypted VPN or direct interconnects.
- Enables Cloud Bursting: baseline sensitive operations run in the private cloud, while temporary seasonal traffic spikes burst into the public cloud.
- Community Cloud:
- Shared collaboratively among multiple organizations with shared missions, security requirements, and regulatory standards (e.g., multi-hospital research consortiums).
5. Virtualization Technologies & Containerization
Virtualization is the foundational enabling technology of modern cloud computing. It abstracts physical computer hardware, allowing a single physical computer (Host) to run multiple isolated, concurrent virtual operating systems (Guests).
TYPE 1 HYPERVISOR (Bare-Metal) TYPE 2 HYPERVISOR (Hosted)
+------------------------------------------+ +------------------------------------------+
| [ Guest VM 1 ] [ Guest VM 2 ] | | [ Guest VM 1 ] [ Guest VM 2 ] |
| (App + OS) (App + OS) | | (App + OS) (App + OS) |
+------------------------------------------+ +------------------------------------------+
| TYPE 1 HYPERVISOR (ESXi, KVM) | | TYPE 2 HYPERVISOR (VirtualBox) |
+------------------------------------------+ +------------------------------------------+
| PHYSICAL HARDWARE | | HOST OPERATING SYSTEM |
| (CPU, RAM, Disks) | | (Windows, macOS, Ubuntu) |
| | +------------------------------------------+
| | | PHYSICAL HARDWARE |
+------------------------------------------+ +------------------------------------------+
5.1 Hypervisors: Type 1 vs. Type 2
| Comparison Dimension | Type 1 Hypervisor (Bare-Metal / Native) | Type 2 Hypervisor (Hosted) |
|---|---|---|
| Installation Layer | Installed directly on physical hardware bare-metal without any underlying host OS. | Installed as an application program on top of an existing host OS (e.g., on Windows 11). |
| Performance Overhead | Near-native physical silicon performance; zero host OS overhead. | Significant latency overhead due to passing calls through the host operating system. |
| Security & Stability | High security; minimal attack surface (no bloated host OS). | Vulnerabilities in host OS directly jeopardize all guest virtual machines. |
| Primary Deployment | Enterprise cloud data centers, AWS servers, corporate virtualization. | Individual developer workstations, academic labs, software testing. |
| Industry Examples | VMware ESXi, Microsoft Hyper-V, KVM (Kernel-based Virtual Machine), Xen. | Oracle VM VirtualBox, VMware Workstation, Parallels Desktop. |
5.2 Virtual Machines (VMs) vs. Containers (Docker)
VIRTUAL MACHINE ARCHITECTURE DOCKER CONTAINER ARCHITECTURE
+------------------------------------------+ +------------------------------------------+
| App 1 | App 2 | App 3 | | App 1 | App 2 | App 3 |
+------------------------------------------+ +------------------------------------------+
| Guest OS 1 | Guest OS 2 | Guest OS 3 | | Bins & Libraries (Packaged dependencies) |
+------------------------------------------+ +------------------------------------------+
| HYPERVISOR | | DOCKER ENGINE |
+------------------------------------------+ +------------------------------------------+
| HOST OPERATING SYSTEM | | HOST OPERATING SYSTEM KERNEL |
+------------------------------------------+ +------------------------------------------+
| HARDWARE | | HARDWARE |
+------------------------------------------+ +------------------------------------------+
Detailed Engineering Comparison
| Feature | Virtual Machine (VM) | Container (Docker) |
|---|---|---|
| Isolation Level | Hardware-level virtualization. Completely isolates guest OS with dedicated virtual hardware. | OS-level virtualization. Shares the host OS kernel; isolates user-space processes. |
| Guest OS Requirement | Mandatory full Guest OS (e.g., full Windows or Ubuntu installation inside VM). | No Guest OS. Packages only application code and required libraries. |
| Storage Footprint | Heavy: $5\text{ to }30\text{ GB}$ per virtual machine image. | Extremely lightweight: $10\text{ to }500\text{ MB}$ per container image. |
| Startup / Boot Time | Slow: $30\text{ seconds to }2\text{ minutes}$ to boot complete guest operating system. | Near-instantaneous: Milliseconds to seconds to start container process. |
| System Resource Usage | High memory and CPU overhead (each VM allocates fixed RAM blocks). | Low overhead; dynamically shares host CPU cores and memory pools. |
| Portability | VM disk images are bulky and difficult to migrate across hypervisors. | Highly portable: Docker images run identically on any machine with Docker Engine. |
6. Whole-Course FCS Master Revision Guide (Lectures 1–45)
| Unit | Primary Modules & Core Topics | Key Focus Areas for Examination |
|---|---|---|
| Unit 1 | Computer Fundamentals, Generations I–V, Von Neumann, Buses, Memory Hierarchy | Characteristics, Von Neumann bottleneck, 5 functional units, bus width calculations. |
| Unit 2 | SRAM vs DRAM, ROM types, HDD vs SSD vs NVMe, Translators (Compiler vs Interpreter) | Flip-flops vs capacitors, seek time + rotational latency, Compiler vs Interpreter table, UEFI. |
| Unit 3 | OS Types, Linux CLI, Octal Permissions, System Calls (fork), Remote SSH |
Batch vs Time-Sharing vs RTOS, chmod 755 math, fork() return values, zombie vs orphan. |
| Unit 4 | Software Crisis, SDLC Phases, Waterfall vs Agile, Git/GitHub Version Control | SRS, 6 SDLC phases, Waterfall vs Scrum comparison, Git 3-tree architecture, merge conflicts. |
| Unit 5 | Cluster, Grid, Edge, Cloud Models (IaaS, PaaS, SaaS), Type 1/2 Hypervisors, VMs vs Containers | 5 NIST cloud characteristics, IaaS vs PaaS vs SaaS, Type 1 vs Type 2 hypervisors, VM vs Docker. |