Cloud Computing (AL-604 (A)) - Unit 3 Short Notes

I. FOUNDATIONAL CONCEPTS & COMPARISONS

A. Core Definition & Characteristics of Cloud Computing

• Definition: A model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, services) that can be rapidly provisioned and released with minimal management effort. (NIST SP 800-145)

• Essential Characteristics:

  1. On-demand self-service: Automatic provisioning without human interaction.

  2. Broad network access: Available via standard mechanisms (e.g., HTTP, mobile).

  3. Resource pooling: Multi-tenant model with dynamic assignment of physical/virtual resources.

  4. Rapid elasticity: Capabilities scale elastically (in/out, up/down) based on demand.

  5. Measured service: Resource usage monitored, controlled, and billed transparently.

B. Computing on Demand (Utility Computing)

• Definition: Treats computing resources (CPU, storage, bandwidth) as a metered utility (like electricity), where users pay only for consumed capacity.

• Dynamic Provisioning:

  • Resources allocated/deallocated in real-time via orchestration tools.

  • Enables elastic scalability: vertical (scale up/down) and horizontal (scale out/in).

  • Automates lifecycle management (provisioning, monitoring, decommissioning).

[!TIP] Utility computing is the business model (pay-per-use); cloud computing is the enabling technology. Often conflated but distinct.

C. Grid Computing vs. Cloud Computing

• Grid Computing:

  • Goal: Solve large, collaborative problems by pooling geographically dispersed, heterogeneous resources.

  • Architecture: Decentralized, with resources contributed by multiple organizations.

  • Example: SETI@home, CERN LHC Computing Grid.

• Cloud Computing:

  • Goal: Deliver standardized, on-demand IT services via centralized data centers.

  • Architecture: Centralized, homogeneous resources in provider-owned data centers.

  • Example: AWS EC2, Microsoft Azure.

• Challenges & Risks:

  • Business Perspective:

    • Grid: High coordination cost, unclear ROI.

    • Cloud: Vendor lock-in, data sovereignty, hidden costs.

  • IT Perspective:

    • Grid: Heterogeneity, fault tolerance, scheduling complexity.

    • Cloud: Multi-tenancy isolation, performance variability, compliance.

[!TIP] Grid = "many computers for one problem"; Cloud = "one provider for many customers". Classic exam distinction.

II. CLOUD DEPLOYMENT MODELS

A. Public Cloud

• Owned/operated by third-party providers (AWS, Azure, GCP).

• Multi-tenant, shared infrastructure, pay-as-you-go.

• Pros: Cost-effective, scalable, no maintenance.

• Cons: Less control, security concerns, compliance limitations.

B. Private Cloud

• Dedicated to a single organization (on-premises or hosted).

• Single-tenant, higher control and security.

• Pros: Customizable, compliant with strict regulations.

• Cons: High capital/operational costs, limited scalability.

C. Hybrid Cloud

• Combines public and private clouds with orchestration for workload mobility.

• Use Cases: Bursting to public cloud during peaks, keeping sensitive data private.

• Pros: Flexibility, optimized cost/security.

• Cons: Complexity in management, integration challenges.

D. Community Cloud

• Shared by organizations with common concerns (e.g., government, healthcare).

• Owned/managed by community or third party.

• Pros: Cost-sharing, industry-specific compliance.

• Cons: Limited scalability, potential conflicts among members.

E. Selection Criteria for Deployment Models

[!TIP] Exam often asks to justify choice based on factors like data sensitivity, budget, scalability, and regulatory needs.

III. CLOUD SERVICE MODELS

A. Software as a Service (SaaS)

• Definition: Applications delivered over the internet, accessed via web browser.

• Provider Responsibilities: Everything (app, data, runtime, middleware, OS, virtualization, servers, storage, networking).

• Examples: Gmail, Salesforce, Office 365.

• Pros: No installation/maintenance, automatic updates, accessible anywhere.

• Cons: Limited customization, data security concerns, dependency on provider.

B. Platform as a Service (PaaS)

• Definition: Platform/environment for developing, testing, and deploying applications.

• Provider Responsibilities: Runtime, middleware, OS, virtualization, servers, storage, networking.

• User Responsibilities: Application and data.

• Examples: Heroku, Google App Engine, AWS Elastic Beanstalk.

• Pros: Faster development, built-in scalability, integrated tools.

• Cons: Limited control over underlying infrastructure, vendor lock-in.

C. Infrastructure as a Service (IaaS)

• Definition: Virtualized computing resources (VMs, storage, networks) on-demand.

• Provider Responsibilities: Virtualization, servers, storage, networking.

• User Responsibilities: OS, middleware, runtime, application, data.

• Examples: AWS EC2, Azure VMs, Google Compute Engine.

• Pros: Maximum control, flexible, pay-per-use.

• Cons: Management overhead, security responsibility on user.

D. Comparative Analysis of SaaS, PaaS, IaaS

E. Specialized Service Concepts

• Storage Cloud:

  • Cloud-based storage services (object, block, file).

  • Examples: Amazon S3 (object), EBS (block), EFS (file).

  • Characteristics: High durability (99.999999999%), scalability, API-driven, pay-per-use.

• Utility Computing as a Service Model:

  • Provision of computing resources (CPU, memory) as a metered utility.

  • Often synonymous with IaaS, emphasizing billing based on consumption (e.g., AWS EC2 on-demand instances).

  • Enables cost optimization by scaling resources dynamically.

IV. VIRTUALIZATION TECHNOLOGY

A. Fundamentals of Virtualization

• Hardware Abstraction: Virtualization layer (hypervisor) abstracts physical hardware, presenting virtual resources (vCPU, vRAM, vDisk) to VMs.

• Resource Pooling: Aggregates physical resources (CPU, memory, storage, network) into shared pools, improving utilization (typically 60–80% vs. 15–20% without virtualization).

• Isolation: Each VM runs in a sandbox, enhancing security and reliability.

B. Hypervisors (Virtual Machine Monitors)

• Definition: Software that creates and manages VMs by virtualizing underlying hardware.

• Functions:

  1. Resource allocation and scheduling.

  2. Isolation between VMs.

  3. Hardware emulation.

  4. VM lifecycle management (create, start, stop, migrate).

• Types:

  | Type 1 (Bare-metal) | Type 2 (Hosted) | |-------------------------|---------------------| | Runs directly on hardware | Runs on top of host OS | | Higher performance, security | Easier setup, lower performance | | Examples: VMware ESXi, Microsoft Hyper-V, Xen | Examples: VMware Workstation, Oracle VirtualBox |

• HVM (Hardware Virtual Machine): Type 1 hypervisor using CPU virtualization extensions (Intel VT-x, AMD-V) for near-native performance.

C. Storage Virtualization

• Concepts & Architecture: Abstracts physical storage devices into logical pools, presented as virtual disks. Virtualization layer sits between hosts and physical storage.

• SAN vs. NAS:

  | Aspect | SAN (Storage Area Network) | NAS (Network-Attached Storage) | |------------------|--------------------------------|-----------------------------------| | Access Level | Block-level (SCSI over FC/iSCSI) | File-level (NFS/CIFS) | | Protocol | Fibre Channel, iSCSI | TCP/IP (Ethernet) | | Performance | High, low latency | Moderate, higher latency | | Use Case | Databases, VM disks (block) | File sharing, backups | | Example | EMC VMAX, NetApp FAS | Synology, QNAP |

D. Logical Partitioning (LPAR)

• Definition: Divides a physical server into isolated logical partitions, each with dedicated resources (CPU, memory, I/O).

• Working: Firmware (e.g., IBM PowerVM) creates partitions; each runs its own OS.

• Advantages:

  • Improved resource allocation and utilization.

  • Strong isolation between partitions.

  • Supports multiple OS on same physical server.

• Disadvantages:

  • Overhead of partition management.

  • Limited scalability vs. full virtualization.

  • Requires specialized hardware/firmware.

[!TIP] LPAR is common in IBM Power Systems; differs from hypervisor-based virtualization (more flexible, less overhead).

E. Virtualized Data Center

• Architecture Components:

  1. Virtualized Servers: Multiple VMs on physical hosts via hypervisor.

  2. Virtualized Storage: Storage pools (SAN/NAS) presented as virtual disks.

  3. Virtualized Networking: Software-defined networking (SDN) for virtual switches, VLANs.

  4. Management Layer: Orchestration tools (e.g., vCenter, OpenStack) for automation.

F. Requirements for Virtualization Platform

• Hardware: CPU virtualization extensions (Intel VT-x/AMD-V), sufficient RAM, I/O virtualization support.

• Hypervisor: Stable, secure, with management interfaces.

• Network: High bandwidth, low latency, virtual networking support.

• Storage: Fast, redundant (RAID), shared storage for VM mobility.

• Management Tools: Monitoring, provisioning, backup solutions.

V. CLOUD SECURITY

A. Importance & Paramount Challenges

• Importance: Multi-tenancy, data sovereignty, compliance (GDPR, HIPAA), trust.

• Challenges:

  • Technical: Data breaches, insecure APIs, account hijacking, insider threats.

  • Business: Vendor lock-in, loss of control, compliance complexity, shared technology risks.

B. Multi-Faceted Cloud Security Framework

1. Data Security: Encryption (at rest/in transit), DLP, data segregation.

2. Network Security: Firewalls, IDS/IPS, DDoS protection, network segmentation.

3. Identity and Access Management (IAM): Authentication (MFA), authorization (RBAC), SSO.

4. Service & Compliance: SLAs, audit trails, certifications (ISO 27001, SOC 2).

C. Secure Execution & Communication

• Encryption: AES for data, TLS/SSL for communication.

• Secure Protocols: HTTPS, SSH, IPsec.

• Secure Bootstrapping: TPM, measured boot, attestation (verify integrity at launch).

D. Virtual Machine (VM) Security

• Risks: VM escape, VM hopping, resource exhaustion, snapshot attacks.

• Best Practices & Recommendations:

  • Use hardened images (minimal OS, patched).

  • Implement network segmentation (micro-segmentation).

  • Continuous VM activity monitoring (logging, anomaly detection).

  • Regular patching and vulnerability scanning.

E. Access Control Mechanisms

• Role-Based Access Control (RBAC):

  • Model: Permissions assigned to roles; users assigned to roles.

  • Components: Users, Roles, Permissions, Sessions.

  • Benefits: Simplifies management, enforces least privilege.

  • Implementation:

    
    Define roles (e.g., Admin, Developer, Viewer).
    
    Assign permissions per role (e.g., Admin: full access; Developer: deploy only).
    
    Assign users to roles.
    
    

  • Example: AWS IAM role "S3ReadOnly" with s3:GetObject permission, assigned to developers.

[!TIP] RBAC is a high-frequency topic. Contrast with ABAC (attribute-based) and MAC (mandatory). Always explain with a cloud example (e.g., AWS IAM).

VI. CLOUD ARCHITECTURE & SERVICE INTEGRATION

A. Cloud Computing Reference Model

• Components/Layers (NIST model):

  1. SaaS: Applications (e.g., Gmail, Salesforce).

  2. PaaS: Development platforms (e.g., Heroku, Google App Engine).

  3. IaaS: Virtualized hardware (e.g., EC2, VMs).

  4. Orchestration Layer: Manages provisioning, scaling, monitoring (e.g., Kubernetes, CloudFormation).

• Diagram: Typically a stack with SaaS on top, PaaS below, IaaS at bottom, orchestration spanning all.

  DiagramSEARCH: NIST cloud computing reference model diagram

B. Service-Oriented Architecture (SOA) in Cloud

• Role: Enables loose coupling between services via standardized interfaces (APIs), facilitating interoperability and integration across cloud platforms.

• Practical Examples:

  • Microservices: Independent services (e.g., user auth, payment) deployed on different clouds, communicating via REST APIs.

  • API Gateways: Aggregate multiple cloud services into a unified interface (e.g., Amazon API Gateway).

  • Integration Platforms: MuleSoft, Apache Camel connect SaaS applications (e.g., Salesforce + SAP).

C. Ecosystem & Stakeholders

• Independent Software Vendors (ISVs):

  • Develop software applications that run on cloud platforms.

  • In e-Business: Build SaaS applications (CRM, ERP) deployed on clouds (AWS, Azure).

  • Role: Leverage cloud scalability for global delivery, reduce infrastructure costs, focus on innovation.

VII. CLOUD MANAGEMENT, PERFORMANCE & TOOLS

A. Quality of Service (QoS)

• Definition: Measurable service attributes defining service level (e.g., availability, reliability, performance).

• Key Metrics:

  • Availability: Uptime percentage (e.g., 99.9% = ~8.76 hours downtime/year).

  • Reliability: Mean Time Between Failures (MTBF).

  • Performance: Response time, throughput, latency.

  • Scalability: Ability to handle load increase.

• Issues:

  • Multi-tenancy causing "noisy neighbor" effect.

  • Dynamic resource allocation affecting consistency.

  • Measuring QoS in distributed, heterogeneous environments.

B. Cloud Infrastructure Benchmarks

• Purpose: Evaluate performance, scalability, efficiency of cloud platforms.

• Metrics:

  • Compute: Instructions per second (IPS), FLOPS.

  • Storage: IOPS (Input/Output Operations Per Second), throughput (MB/s).

  • Network: Bandwidth, latency, jitter.

• Standards: SPEC Cloud, TPC-C (transaction processing), YCSB (NoSQL).

C. Cloud Management Platforms & Tools

• OpenNebula:

  • Architecture: Front-end (Sunstone GUI, CLI), back-end (scheduler, monitoring, storage/network drivers).

  • Usage: Open-source platform for building private/public/hybrid clouds. Manages VMs, networks, storage via hypervisors (KVM, VMware).

• Nimbus:

  • Functionality: Toolkit for IaaS (VM management, context customization).

  • Role: Used in scientific/HPC clouds (e.g., for research grids). Integrates with OpenStack.

D. OLAP in Cloud Context

• Functionality: Online Analytical Processing for multidimensional data analysis (business intelligence).

• Operations:

  1. Roll-up: Aggregate data (e.g., sum sales by region).

  2. Drill-down: View detailed data (e.g., sales by city).

  3. Slice: Select one dimension (e.g., sales in 2023).

  4. Dice: Select sub-cube (e.g., sales in 2023 for product X).

  5. Pivot (Rotate): Reorient cube (e.g., swap rows/columns).

• Cloud Benefit: Scalable storage/compute for large datasets, on-demand analytics.

[!TIP] OLAP operations are frequently asked with examples. Memorize the five operations and their purposes.