Skip to content
Rounder
Start
BCA-304N · Cloud Computing
Unit 1

Introduction to Parallel, Distributed and Cloud Computing

Sign in to save
Unit 1 Learning Hub

Introduction to Parallel, Distributed and Cloud Computing

Core Definitions & Key Concepts

Core Definitions & Key Concepts

Cloud Computing according to NIST (National Institute of Standards and Technology)

NIST defines Cloud Computing as a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.

What are the 5 essential characteristics of Cloud Computing defined by NIST

  1. On-Demand Self-Service
  2. Broad Network Access
  3. Resource Pooling (Multi-tenancy)
  4. Rapid Elasticity
  5. Measured Service (Pay-as-you-go)

Parallel Computing and Distributed Computing

  • Parallel Computing: Simultaneous execution of computational tasks across multiple processing cores sharing centralized physical memory or low-latency interconnects.
  • Distributed Computing: Autonomous computational systems located on distinct networked physical machines coordinating actions by message passing, without shared physical memory.

What is Utility Computing

A service provisioning model where computational infrastructure, bandwidth, and storage are packaged and billed as a metered service, analogous to traditional public utilities (water, electricity, telephone).

Grid Computing and how it served as a precursor to the Cloud

Grid Computing pools geographically distributed, heterogeneous, autonomous supercomputers and cluster resources across administrative boundaries to solve monolithic scientific workloads (e.g., CERN particle simulations). Cloud evolved Grid towards dynamic multi-tenant commercial virtualization.

What role did Web 2.0 technologies play in the evolution of Cloud Computing

Web 2.0 shifted the web from static read-only HTML pages into interactive, dynamic, rich-internet web applications (AJAX, REST APIs, JSON), enabling browser-based client interfaces to orchestrate backend cloud services seamlessly.

The Cloud Computing Reference Model layers

  1. Infrastructure Layer (IaaS): Raw compute VMs, storage blocks, and virtual networks.
  2. Platform Layer (PaaS): Development runtimes, databases, execution environments.
  3. Application Layer (SaaS): Finished end-user software delivered over the web.

What is Multi-Tenancy in cloud computing architectures

A software architecture where a single instance of a software application or physical infrastructure serves multiple independent consumer organizations ('tenants'), isolating their data securely via logical virtualization.

What is Service-Oriented Computing (SOC) and Service-Oriented Architecture (SOA)

A design paradigm where modular software functionality is partitioned into loosely coupled, interoperable, platform-independent services that communicate over standard network protocols (e.g., SOAP/XML, REST/JSON).

Autonomic Computing and its significance in cloud data centers

IBM's concept of self-managing computing systems that dynamically regulate their own behavior through a MAPE-K loop (Monitor, Analyze, Plan, Execute, over shared Knowledge) to self-heal, self-configure, self-optimize, and self-protect.

✓Topic 1 covered: Core Definitions & Key Concepts
Next: Conceptual & Architectural Analysis↓

Conceptual & Architectural Analysis

Conceptual & Architectural Analysis

Compare Distributed Systems, Grid Computing, Cluster Computing, and Cloud Computing

DimensionCluster ComputingGrid ComputingCloud Computing
HomogeneityHighly homogeneous nodesHeterogeneous, dispersedHeterogeneous hardware, uniform VM abstraction
Resource OwnershipSingle centralized entityMultiple administrative domainsCloud Service Provider (CSP)
Primary WorkloadHigh Performance Computing (HPC)Monolithic batch scientific runsDynamic web apps, multi-tenant enterprise
VirtualizationOptional (Bare metal preferred)Rare / MinimalMandatory (Core abstraction layer)
Billing ModelFixed capital expenditure (CapEx)Collaborative / Grant-fundedOpEx (Pay-as-you-go, per second billing)

The NIST Cloud Computing Reference Architecture with Block Diagrams

NIST Cloud Conceptual Roles

System Architecture / Concept DiagramHorizontal Scroll Enabled
+-------------------------------------------------------------------------+
|                           CLOUD CONSUMER                                |
|             (SaaS Consumer / PaaS Developer / IaaS Admin)               |
+------------------------------------+------------------------------------+
                                     | Consumes Services
                                     v
+-------------------------------------------------------------------------+
|                           CLOUD PROVIDER                                |
|  +--------------------+  +--------------------+  +--------------------+ |
|  |   SaaS Services    |  |   PaaS Runtimes    |  |   IaaS Resources   | |
|  +--------------------+  +--------------------+  +--------------------+ |
|  | Service Management: Business / Provisioning / Portability / SLA    | |
|  | Security & Privacy: Identity Access Management (IAM), Encryption   | |
+------------------------------------+------------------------------------+
        |                                              ^
        | Coordinates Audit                            | Brokered Mediation
        v                                              |
+-----------------------+                    +----------------------------+
|     CLOUD AUDITOR     |                    |        CLOUD BROKER        |
| Security / Performance|                    | Intermediation / Arbitrage |
+-----------------------+                    +----------------------------+
  • Cloud Consumer: End user or enterprise entity browsing or deploying workloads.
  • Cloud Provider: Entity responsible for acquiring hardware, facilities, and delivering services.
  • Cloud Auditor: Independent third-party assessing security, privacy, and SLA adherence.
  • Cloud Broker: Entity managing relationship, aggregation, and integration between multiple CSPs.
  • Cloud Carrier: Telecommunications intermediary providing network transport connectivity.

Examine the Historical Timeline and Evolutionary Drivers of Cloud Computing

The Evolutionary Roadmap of Computing

  1. Mainframe Era (1960s): Centralized time-sharing computing on IBM mainframes; dumb terminal clients.
  2. Personal Computer (PC) Revolution (1980s): Decentralized client processing; corporate servers handle file and print sharing.
  3. Client-Server & Distributed Systems (1990s): Multi-tier applications, RPC, CORBA, cluster computing for fault tolerance.
  4. Dot-Com Era & Web 2.0 (early 2000s): High-speed broadband, XML-RPC, dynamic web applications.
  5. Commercial Cloud Dawn (2006): Amazon launches S3 and EC2, proving commoditized infrastructure on demand.
  6. Modern Cloud Native & Serverless (2014–Present): Docker containers, Kubernetes orchestration, FaaS (AWS Lambda), multi-cloud enterprise.

Detail the Five Essential Characteristics of Cloud Computing with Concrete Scenarios

  1. On-Demand Self-Service: A developer provisions an 8-core Ubuntu VM with 32GB RAM via AWS Management Console in 45 seconds without contacting a human system administrator.
  2. Broad Network Access: Resources are accessible over HTTPS from heterogeneous client platforms (smartphones, thin client terminals, laptops) globally.
  3. Resource Pooling: CSP dynamically assigns physical CPU cores, RAM, and NVMe drives across multiple multinational companies, isolating them using hypervisor hardware virtualization.
  4. Rapid Elasticity: An e-commerce platform automatically scales from 4 EC2 instances to 80 instances during a Black Friday flash sale and scales back down when traffic subsides.
  5. Measured Service: Metering systems track compute seconds, gigabytes transferred, and I/O operations, invoicing the tenant precisely for resource units consumed.

Service-Oriented Architecture (SOA) Principles and Loose Coupling in Cloud Computing

Principles of SOA in Cloud

  • Loose Coupling: Services minimize dependencies on each other; consumers interact strictly via formal public contracts (WSDL, OpenAPI/Swagger), allowing internal implementations to change without breaking clients.
  • Service Reusability: Business logic is encapsulated into atomic, composable enterprise services (e.g., authentication, credit processing).
  • Statelessness: Services maintain no transactional state between requests; session states are externalized to distributed caches (Redis), enabling horizontal scaling.
  • Discoverability: Services register metadata with service registries (Consul, Eureka) allowing dynamic load balancing and endpoint routing.
✓Topic 2 covered: Conceptual & Architectural Analysis
Next: In-Depth Analysis & Implementation↓

In-Depth Analysis & Implementation

In-Depth Analysis & Implementation

Analyze the Architectural Hierarchy of Cloud Computing (IaaS vs PaaS vs SaaS), Specifying Customer vs Provider Responsibility Split (The Shared Responsibility Model)

The Cloud Responsibility Continuum

System Architecture / Concept DiagramHorizontal Scroll Enabled
+-----------------------+-----------+-----------+-----------+-----------+
| Architectural Layer   | On-Prem   |   IaaS    |   PaaS    |   SaaS    |
+-----------------------+-----------+-----------+-----------+-----------+
| Applications          | Customer  | Customer  | Customer  | Provider  |
| Data & Content        | Customer  | Customer  | Customer  | Provider  |
| Runtime & Execution   | Customer  | Customer  | Provider  | Provider  |
| Middleware            | Customer  | Customer  | Provider  | Provider  |
| Operating System (OS) | Customer  | Customer  | Provider  | Provider  |
| Hypervisor            | Customer  | Provider  | Provider  | Provider  |
| Physical Servers      | Customer  | Provider  | Provider  | Provider  |
| Storage Hardware      | Customer  | Provider  | Provider  | Provider  |
| Physical Networking   | Customer  | Provider  | Provider  | Provider  |
| Facility Security     | Customer  | Provider  | Provider  | Provider  |
+-----------------------+-----------+-----------+-----------+-----------+

1. Infrastructure as a Service (IaaS)

  • Provider Responsibility: Data center facilities, power, physical cooling, bare-metal servers, physical switches, storage arrays, and hypervisor virtualization layer.
  • Consumer Responsibility: Guest operating system (installing patches, configuring kernel parameters), network firewall rules (Security Groups), middleware, runtimes, database instances, and applications.
  • Representative Platforms: Amazon EC2, Google Compute Engine (GCE), Microsoft Azure Virtual Machines.

2. Platform as a Service (PaaS)

  • Provider Responsibility: Hardware through OS, language runtime engines (Java, Python, Node.js), middleware, automated patching, scaling triggers, and database engine maintenance.
  • Consumer Responsibility: Application source code, schema architecture, and application-level configuration.
  • Representative Platforms: Google App Engine, AWS Elastic Beanstalk, Heroku.

3. Software as a Service (SaaS)

  • Provider Responsibility: Complete end-to-end management from physical silicon through application code, bug fixing, high availability, backup redundancy, and disaster recovery.
  • Consumer Responsibility: User credential governance, identity access policies (RBAC), and customer data entry.
  • Representative Platforms: Google Workspace, Microsoft 365, Salesforce CRM, Dropbox.

Design and Detail the Complete Architecture of a Modern Cloud Data Center Fabric (Leaf-Spine Network, Storage Fabrics, Power & Cooling)

Hyperscale Cloud Data Center Architecture

System Architecture / Concept DiagramHorizontal Scroll Enabled
                       [ Spine Switches (Layer 3) ]
                       /      |              |      \
                      /       |              |       \
         [ Leaf Switch 1 ]  [ Leaf Switch 2 ]  [ Leaf Switch 3 ]  (ToR Switches)
             /     \            /     \            /     \
       [ Server 1 ] [ S2 ] [ Server 3 ] [ S4 ] [ Server 5 ] [ S6 ]

1. Leaf-Spine (Clos) Network Topology

  • Traditional three-tier (Core-Aggregation-Access) tree architectures choke on modern cloud East-West traffic (server-to-server traffic for distributed databases and microservices).
  • Leaf-Spine Fabric: Every Leaf switch (Top-of-Rack - ToR) is directly connected to every Spine switch. Guarantees non-blocking, deterministic maximum 2-hop latency between any two arbitrary servers in the data center, leveraging Equal-Cost Multi-Path (ECMP) routing.

2. Storage Area Network (SAN) & Distributed Object Stores

  • NVMe-over-Fabrics (NVMe-oF) leveraging RDMA (Remote Direct Memory Access) over Converged Ethernet (RoCE), delivering sub-millisecond block latency.
  • Erasure Coding (e.g., Reed-Solomon 8+4) replaces raw triple replication, slashing storage overhead from 300% to 150% while surviving concurrent rack failures.

3. Power, PUE, and Thermal Engineering

  • Power Usage Effectiveness (PUE):

$PUE = \frac{Total Facility Power Ingestion}{Total IT Equipment Energy Consumption}$ Industry average is 1.5–1.8; modern hyperscalers achieve 1.1–1.15 using adiabatic evaporative cooling, cold-aisle containment, and direct-to-chip liquid cooling.

Examine the Economic and Operational Paradigm Shift: Capital Expenditure (CapEx) vs Operational Expenditure (OpEx) in Cloud Transformations

Cloud Financial Operations (FinOps) & Economic Models

1. Traditional On-Premise CapEx Model

  • Involves massive upfront capital investment in physical servers, redundant diesel generators, air chillers, and software enterprise licenses with 3–5 year amortization lifecycles.
  • Over-Provisioning Trap: System architects must provision hardware to withstand peak holiday traffic spikes. For 95% of the year, up to 80% of server capacity sits completely idle, consuming idle power.

2. Cloud Utility OpEx Model

  • Eliminates upfront capital expenditure, converting infrastructure into variable operational expenses paid out of ongoing operational budgets.
  • Elasticity Curve Alignment: Cloud auto-scaling dynamically adjusts provisioned capacity to match real-time demand, completely eliminating the deadweight loss of idle over-provisioning.

3. Total Cost of Ownership (TCO) Decomposition

$TCO_{On-Prem} = Hardware + Software Licenses + Real Estate + Electricity/Cooling + Network Carrier Bandwidth + SysAdmin Labor$ While cloud raw per-hour compute costs may appear higher than amortized silicon, TCO analysis demonstrates significant savings through zero facility maintenance, zero real-estate footprint, and automated self-healing operations.

✓Topic 3 covered: In-Depth Analysis & Implementation
Next: Important Exam Points & Quick Revision↓

Important Exam Points & Quick Revision

Important Exam Points & Quick Revision

Examiners heavily test: NIST 5 essential characteristics (5 marks), IaaS vs PaaS vs SaaS shared responsibility matrix (10/15 marks), and CapEx vs OpEx economic transition (5 marks).

Practice Ready

Ready to practice Unit 1?

You've read the core concepts. Take the balanced 10-question practice set to test your mastery!

Save progress