Skip to content
CampusPrep
IT-702 (B) · Cloud Computing/Quick Revision Short Notes

Cloud Computing (IT-702 (B)) - Unit 3 Short Notes

I. CLOUD COMPUTING FUNDAMENTALS

Definition: Cloud computing is 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 or service provider interaction.

Essential Characteristics (NIST Definition):

  1. On-demand self-service: Users can provision computing resources automatically without human interaction.

  2. Broad network access: Resources are available over the network via standard mechanisms (e.g., HTTP, HTTPS).

  3. Resource pooling: Provider's computing resources are pooled to serve multiple consumers (multi-tenant model).

  4. Rapid elasticity: Resources can be elastically provisioned and released to scale rapidly with demand.

  5. Measured service: Resource usage is monitored, controlled, and reported, enabling pay-per-use billing.

Service Models:

Model Provides User Control Example
IaaS Virtualized hardware (VMs, storage, networks) OS, apps, data AWS EC2, Azure VMs
PaaS Runtime environment, middleware, development tools Apps & data Google App Engine, Heroku
SaaS Complete applications accessible via web browser App configuration only Gmail, Salesforce

Deployment Models:

  • Public Cloud: Owned/operated by third-party providers (AWS, Azure). Multi-tenant, pay-as-you-go.

  • Private Cloud: Exclusive use by a single organization. Can be on-premises or hosted. Offers more control.

  • Hybrid Cloud: Composition of two or more clouds (private/public) with orchestration between them.

  • Community Cloud: Shared by several organizations with common concerns (security, compliance). (See Unit VII for details).

Key Benefits:

  • Reduced Time to Market: Rapid provisioning accelerates development and deployment cycles.

  • Lower Capital Expenses (CapEx): Converts fixed infrastructure costs to variable operational expenses (OpEx).

  • Elasticity & Scalability: Instantly scale resources up/down based on real-time demand.

  • Global Scale: Access to a vast, geographically distributed infrastructure.

  • Increased Efficiency & Utilization: Resource pooling leads to higher utilization rates than traditional silos.

[!TIP] Exam Focus: Be prepared to differentiate between IaaS, PaaS, and SaaS based on the level of control and management responsibility. Also, clearly distinguish deployment models by ownership, purpose, and cost model.

II. CLOUD PLATFORMS AND SERVICES

Platform as a Service (PaaS)

  • Essential Characteristics: Provides a platform allowing customers to develop, run, and manage applications without the complexity of building and maintaining infrastructure. Includes OS, programming language execution environment, database, web server.

  • Key Features: Built-in scalability, high availability, integrated middleware/services (messaging, caching), automated provisioning, and often a multi-tenant architecture.

Google App Engine (GAE)

  • Major Features:

    • Fully managed PaaS for building scalable web apps.

    • Supports multiple runtimes (Python, Java, Go, PHP, Node.js).

    • Automatic scaling based on traffic.

    • Built-in services (Datastore, Memcache, Task Queues, Mail API).

    • Pay-per-use pricing model.

  • Types of Problems Solvable: Web applications, mobile backends, APIs, data-driven applications. Ideal for applications with variable or unpredictable load.

Microsoft Azure

  • Virtualization Techniques: Heavily relies on Microsoft Hyper-V as its core hypervisor for creating and managing virtual machines. Uses a network virtualization layer to create virtual networks and subnets, isolating tenant traffic. Employs storage virtualization to abstract physical storage into logical units (LUNs) presented to VMs.

Eucalyptus

  • Features & Modes:

    • Open-source software for building AWS-compatible private/hybrid clouds.

    • Architecture: Components include Cloud Controller (CLC), Cluster Controller (CC), Storage Controller (SC), Node Controller (NC).

    • Modes:

      1. Managed Mode: Eucalyptus manages its own resources (VMs, networking, storage).

      2. Eucalyptus-to-AWS Mode (Hybrid): Workloads can be bursted to AWS public cloud.

      3. AWS-to-Eucalyptus Mode: AWS resources can be managed within the Eucalyptus environment.

    • Provides AWS-compatible APIs (EC2, S3, IAM).

Storage Solutions

  • Storage Cloud: A cloud service model where data is stored on remote storage systems accessed via the internet (e.g., AWS S3, Azure Blob Storage). Key concepts: object storage (flat namespace, metadata), block storage (for VMs), file storage (NAS-like).

  • Storage Area Network (SAN): A dedicated high-speed network that interconnects storage devices (disk arrays) with servers. Provides block-level access, appearing as local disk to the OS. Key technologies: Fibre Channel, iSCSI. Cloud Context: Often used as the backend storage for IaaS cloud providers' block storage offerings (e.g., EBS).

[!TIP] Exam Focus: Know the specific virtualization tech for Azure (Hyper-V) and the AWS compatibility aspect of Eucalyptus. Distinguish SAN (block-level, high-speed network) from NAS (file-level, IP network) and cloud object storage.

III. BIG DATA AND ANALYTICS ON CLOUD

A. Big Data Fundamentals

  • Definition: Extremely large, complex datasets that traditional data processing applications are inadequate to handle. Characterized by the 3 Vs:

    1. Volume: Scale of data (Terabytes to Zettabytes).

    2. Variety: Different forms (structured, semi-structured, unstructured: text, logs, images, video).

    3. Velocity: Speed of data generation, ingestion, and processing (real-time/streaming).

  • Challenges: Storage, processing, analysis, visualization, security, privacy, governance, talent gap.

  • Big Data Analytics: Process of examining large, varied datasets to uncover hidden patterns, correlations, and insights. Includes descriptive, diagnostic, predictive, and prescriptive analytics.

  • Real-World Applications: Fraud detection (finance), personalized recommendations (retail), predictive maintenance (manufacturing), genomics (healthcare), smart cities (IoT), sentiment analysis (social media).

B. Data Preprocessing

  • Data Cleaning: Process of fixing or removing incorrect, incomplete, duplicate, or corrupted data.

    • Techniques: Handling missing values (deletion, imputation), smoothing noise (binning, regression), resolving inconsistencies, outlier detection (IQR, Z-score).

  • Sampling: Technique to select a representative subset of data for analysis when full dataset is too large.

    • Methods: Simple random sampling, stratified sampling (proportional from subgroups), systematic sampling, cluster sampling.

C. Machine Learning for Data Analysis

  • Classification Algorithms:

    • Decision Trees: Supervised learning method that predicts target variable by learning simple decision rules from data features.

      • Types & Explanation:

        • ID3: Uses Information Gain (based on entropy reduction) to select splits. Handles only categorical features.

        • C4.5: Extension of ID3. Uses Gain Ratio (normalized Information Gain) to handle bias towards multi-valued attributes. Handles continuous features and missing values.

        • CART (Classification and Regression Trees): Uses Gini Impurity for classification. Generates binary splits. Can also perform regression.

      • Process: Recursive partitioning: select best attribute to split on, create branches, repeat until stopping condition (pure nodes, max depth, min samples).

    • Naive Bayes Classification: Based on Bayes' Theorem with "naive" assumption of feature independence given the class.

      • Principle:

$$P(Class|Features) \propto P(Class) \prod_{i=1}^{n} P(Feature_i|Class)$$

    *   Predicts the class with the highest posterior probability. Efficient, works well with high-dimensional data. Sensitive to irrelevant features due to independence assumption.

  • Association Rules: Discovers interesting relations (rules) between variables in large databases.

    • Concept: Rule: X => Y (if X then Y). Measured by:

      • Support: P(X ∪ Y) - frequency of rule.

      • Confidence: P(Y|X) = Support(X∪Y)/Support(X) - predictive power.

      • Lift: Confidence / P(Y) - strength of rule over random chance.

    • Mining Algorithm: Apriori Algorithm: Uses candidate generation and pruning based on minimum support. Finds all frequent itemsets, then generates rules meeting minimum confidence.

    • Applications: Market basket analysis, cross-selling, recommendation systems, medical diagnosis.

D. Programming for Data Analysis: R

  • Features of R:

    • Open-source, interpreted language.

    • Vectorized operations (efficient on arrays).

    • Extensive statistical and graphical capabilities.

    • Large, active package ecosystem (CRAN).

    • Strong data handling and storage capabilities.

    • Cross-platform.

  • Major Components of R Environment:

    • R Console: Interactive command-line interface.

    • R Script Editor: For writing and saving scripts (e.g., RStudio).

    • Workspace: Current objects (variables, data frames) in memory.

    • Package Manager: For installing/loading libraries.

    • Help System: ?function or help(function).

  • Operations on Vectors:

    • Creation: v <- c(1, 2, 3), v <- 1:10, v <- seq(1, 10, by=0.5).

    • Indexing: v[2] (second element), v[c(1,3)], v[v>5] (logical indexing).

    • Arithmetic: Element-wise operations (v1 + v2, v * 2).

    • Recycling Rule: Shorter vector is recycled to match longer vector length in operations.

    • Functions: length(v), sum(v), mean(v), sort(v), v[order(v)].

E. Big Data Processing Frameworks

  • Hadoop Ecosystem Architecture:

    DiagramCANVAS: HDFS Architecture showing Client, NameNode (metadata), DataNodes (block storage), with blocks replicated across DataNodes. YARN architecture showing ResourceManager, NodeManager, ApplicationMaster, Container.

    • HDFS (Hadoop Distributed File System):

      • Master-Slave architecture. NameNode (master) manages metadata (file system tree, block locations). DataNodes (slaves) store actual data blocks (default 128MB/256MB).

      • Files are split into blocks, distributed across DataNodes, and replicated (default 3x) for fault tolerance.

      • Write-once-read-many (WORM) model.

    • YARN (Yet Another Resource Negotiator):

      • Resource management layer. Separates resource management (ResourceManager) from job scheduling/monitoring (ApplicationMaster per application).

      • Components: ResourceManager (scheduler, applications manager), NodeManager (per-node resource/container manager), ApplicationMaster (per-application negotiator).

    • MapReduce Programming Model:

      • Parallel, distributed computation on HDFS data.

      • Phases: Map (processes input key-value pairs, emits intermediate key-value pairs) -> Shuffle & Sort (groups values by key) -> Reduce (aggregates values for each key, emits final output).

      • Example - Word Count:

        • Map: Input (docID, text). Emits (word, 1) for each word.

        • Reduce: Input (word, [1,1,...]). Sums counts: (word, total_count).

      • Types/Formats: Map, Reduce, Combiner (local reducer to reduce network traffic), Partitioner (controls key distribution to reducers). Input/Output formats: TextInputFormat, KeyValueTextInputFormat, SequenceFileInputFormat.

      • Applications: Log processing, indexing, graph processing (e.g., PageRank), ETL.

  • Apache Hive:

    • Architecture Overview:

      • Hive CLI/Web UI: User interface.

      • Driver: Manages lifecycle of a Hive query (compilation, optimization, execution).

      • Compiler: Parses query, generates logical plan (Abstract Syntax Tree), optimizes (logical & physical), creates execution plan (MapReduce/Tez/Spark jobs).

      • Metastore: Stores metadata (table schema, location, partition info) in relational DB (MySQL, Derby).

      • Execution Engine: Executes the plan on Hadoop (YARN).

    • Procedure to Write User-Defined Functions (UDFs):

      1. Extend org.apache.hadoop.hive.ql.exec.UDF class.

      2. Implement public Object evaluate() method with custom logic.

      3. Package as JAR file.

      4. Add JAR to Hive session: ADD JAR <path_to_jar>.

      5. Create temporary/permanent function: CREATE TEMPORARY FUNCTION my_udf AS 'com.example.MyUDF'.

      6. Use in HiveQL queries: SELECT my_udf(column) FROM table;.

  • Apache Pig:

    DiagramCANVAS: Pig Architecture showing Pig Latin script -> Parser -> Logical Plan -> Optimizer (Logical & Physical) -> Physical Plan -> Execution Engine (MapReduce/Tez/Spark).

    • Pig Latin Application Flow:

      1. Write Pig Latin script.

      2. Parse & Validate script syntax.

      3. Logical Plan created (series of logical operators).

      4. Optimize logical plan (combine operations, push filters).

      5. Physical Plan created (MapReduce/Tez/Spark jobs).

      6. Execute plan on Hadoop cluster.

    • Basics of Pig Latin Scripting:

      • Data Loading: A = LOAD 'data' USING PigStorage(',') AS (field1:chararray, field2:int);

      • Transformations: B = FILTER A BY field2 > 10; C = GROUP B BY field1; D = FOREACH C GENERATE group, COUNT(B);

      • Storing: STORE D INTO 'output' USING PigStorage('\t');

[!TIP] Exam Focus: Be able to draw and explain HDFS & YARN architectures. Know the exact MapReduce flow with a clear example (Word Count). For Hive UDFs, remember the Java class extension and CREATE FUNCTION syntax. For Pig, understand the script flow from LOAD to STORE and the difference between GROUP and JOIN.

IV. CLOUD SECURITY AND TRUST

  • Security Challenges:

    • Data Security: Breaches, unauthorized access, data loss, insecure APIs.

    • Privacy: Compliance (GDPR, HIPAA), data location/jurisdiction, monitoring by provider.

    • Compliance: Meeting industry-specific regulations in a shared, dynamic environment.

    • Multi-tenancy Issues: Isolation failures between tenants (noisy neighbor), side-channel attacks.

    • Account Hijacking: Credential theft leading to data manipulation.

    • Insider Threats: Malicious employees at cloud provider.

    • Denial-of-Service (DoS): Overwhelming cloud services.

  • Cloud Computing Security Architecture (Layered Model):

    DiagramCANVAS: Layered diagram from bottom to top: 1. Physical Security (data center), 2. Network Security (firewalls, IDS/IPS, VPN), 3. Host Security (hypervisor hardening, VM security), 4. Application Security (secure coding, WAF), 5. Data Security (encryption at rest/in transit, DLP), 6. Identity & Access Management (IAM, SSO, MFA).
    • Key Components: IAM, encryption (symmetric/asymmetric), key management, security monitoring/logging, incident response, compliance auditing.

  • Trusted Cloud Computing:

    • Concepts: Establishing confidence that the cloud infrastructure and services operate as intended and are free from unauthorized manipulation. Involves trust in hardware, hypervisor, provider, and data.

    • Mechanisms:

      • Hardware Root of Trust: TPM (Trusted Platform Module) for attestation.

      • Attestation: Verifying the integrity of the software stack (hypervisor, VMs) via cryptographic measurements.

      • Secure Multi-tenancy: Strong isolation (VMs, containers), secure resource scheduling.

      • Transparency & Auditing: Regular third-party audits (SOC 2, ISO 27001), audit logs accessible to customers.

      • SLAs with Security Guarantees: Contractual terms for security controls and breach notifications.

V. CLOUD PERFORMANCE AND QUALITY OF SERVICE

  • Quality of Service (QoS) Issues:

    • Challenges: Resource contention in multi-tenant environments, network latency/variability, storage I/O bottlenecks, "noisy neighbor" effect, ensuring consistent performance for latency-sensitive apps, measuring and guaranteeing SLAs.

    • Metrics: Availability (%), Latency (ms), Throughput (requests/sec), Response Time, Error Rate, Jitter.

    • Example: A web application hosted on a public cloud may experience performance degradation during peak hours if the underlying physical host is oversubscribed by other VMs.

  • Elastic Computing:

    • Concepts: Ability to dynamically and automatically scale computing resources (up/down) in response to workload changes, while maintaining performance and meeting SLAs. Core benefit of cloud computing.

    • Importance: Cost optimization (pay for what you use), handling unpredictable spikes (flash crowds), maintaining application responsiveness, disaster recovery readiness.

    • Implementation Strategies:

      • Auto-scaling: Based on metrics (CPU > 80%, network in). Policies: scale-out (add instances), scale-in (remove instances), scale-up/down (change instance size).

      • Load Balancing: Distributes traffic across multiple instances (e.g., round-robin, least connections).

      • Right-sizing: Choosing appropriate instance types for workloads.

      • Scheduled Scaling: Scaling based on known time patterns (e.g., business hours).

[!TIP] Exam Focus: Differentiate between scalability (design to handle growth) and elasticity (automatic, dynamic scaling). Know key QoS metrics and how they are measured. Link elasticity directly to cost savings and performance.

VI. CLOUD APPLICATION DOMAINS

A. Social Networking Applications

  • Advantages of Cloud Technologies:

    • Massive Scalability: Handle millions of concurrent users and explosive data growth (posts, images, videos).

    • Cost-Effectiveness: Pay-per-use model aligns with unpredictable, spiky traffic patterns.

    • Global Reach: CDN integration and multi-region deployment for low-latency access worldwide.

    • Rapid Innovation: Access to PaaS services (DBs, caching, analytics) accelerates feature development.

    • High Availability: Built-in redundancy and failover mechanisms ensure uptime.

B. Simulation and Modeling on Cloud

  • Simulation Fundamentals:

    • Definition: Imitation of the operation of a real-world process or system over time to study its behavior and evaluate strategies.

    • Steps (with Flow Diagram):

      DiagramCANVAS: Flowchart: 1. Problem Identification & Objectives -> 2. Data Collection & Input Modeling -> 3. Conceptual Model Building -> 4. Verification & Validation -> 5. Experimentation & Output Analysis -> 6. Documentation & Implementation.

    • Continuous vs. Discrete System Simulation:

      • Continuous: State variables change continuously over time. Modeled using differential equations (e.g., fluid dynamics, chemical processes).

      • Discrete: State variables change at distinct points in time (events). Modeled using event scheduling/process interaction (e.g., queuing systems, manufacturing lines).

  • Probability and Statistics for Simulation:

    • Distributions:

      • Binomial:

$$P(X=k) = \binom{n}{k} p^k (1-p)^{n-k}$$

, k successes in n trials.

    *   **Poisson:**

$$P(X=k) = \frac{\lambda^k e^{-\lambda}}{k!}$$

, k events in fixed interval.

    *   **Normal:**

$$f(x) = \frac{1}{\sigma\sqrt{2\pi}} e^{-\frac{1}{2}\left(\frac{x-\mu}{\sigma}\right)^2}$$

, continuous, symmetric.

*   **Binomial Approximation by Poisson:** When number of trials `n` is very large (`n → ∞`), probability of success `p` is very small (`p → 0`), and `λ = n*p` is moderate (finite). Conditions: `n ≥ 20` and `p ≤ 0.05` or `n ≥ 100` and `np ≤ 10`.

*   **Stochastic Variables:** Random variables whose values are outcomes of a random phenomenon.

*   **Density & Distribution Functions:** PDF `f(x)` gives relative likelihood; CDF `F(x) = P(X ≤ x)` gives cumulative probability.

*   **Arrival Patterns:** Often modeled as Poisson process (exponential inter-arrival times) for "random" arrivals.

*   **Differential Equations:** Used in continuous simulation to describe system dynamics (e.g., `dx/dt = f(x,t)`).

  • Queuing Theory:

    • General Queuing System (Kendall Notation A/B/c): A = arrival distribution, B = service distribution, c = number of servers.

      DiagramCANVAS: Queue diagram: Arrival Process -> Queue (buffer) -> c parallel Service Channels -> Departure. Key parameters: λ (arrival rate), μ (service rate per channel), ρ = λ/(cμ) (utilization).

    • Simulation of Queuing Systems: Generate random inter-arrival and service times from distributions, track queue length, waiting time, server utilization over simulated time.

    • Applications & Characteristics: Call centers, traffic flow, computer networks, manufacturing. Key metrics: average queue length, average waiting time, probability of delay, server utilization.

  • Simulation Techniques and Validation:

    • Verification vs. Validation:

      • Verification: "Did we build the model right?" (Code debugging, checking logic).

      • Validation: "Did we build the right model?" (Accuracy in representing real system). Methods: face validation, sensitivity analysis, historical data comparison.

    • AI Techniques in Simulation: Neural networks for input modeling or surrogate models, fuzzy logic for handling uncertainty, genetic algorithms for optimization.

    • Analog vs. Digital Simulation:

      • Analog: Uses physical models (e.g., wind tunnel for aerodynamics). Continuous.

      • Digital: Uses computer models (discrete-event, continuous). Most common today.

    • Autopilot Simulation with Example: Simulating aircraft control systems. Example: Simulating a simple altitude hold system where the autopilot adjusts elevator based on error between desired and actual altitude.

    • Pure Pursuit Problem: Path-tracking algorithm for autonomous vehicles. The vehicle steers towards a point a fixed look-ahead distance ahead on the desired path. Used in agricultural robots, self-driving cars.

  • Simulation Languages and Tools:

    • Simulation of Classification Languages: General-purpose languages (C++, Java) with libraries (SimPy for Python). Specialized simulation languages (GPSS, Simscript) offer built-in constructs for entities, queues, resources.

C. Augmented and Virtual Reality on Cloud

  • Virtual Reality (VR):

    • Definition & How it Works: Immersive, computer-generated simulation of a 3D environment. Uses Head-Mounted Display (HMD) with stereoscopic screens, head tracking (sensors, cameras), and input devices (controllers). Tracks user's head/body movements to update view in real-time, creating a sense of presence.

    • Geometric Modeling: Representation of 3D objects. Types:

      • Wireframe: Edges/vertices only. Fast, but ambiguous.

      • Surface: B-rep (boundary representation), polygon meshes. Defines object surface.

      • Solid: Defines volume (CSG - Constructive Solid Geometry, B-rep with volume). Used for physical simulation.

    • Interpolation & Translation:

      • Interpolation: Estimating values between known points. Types: Linear (straight line), Polynomial (smooth curve), Spline (piecewise polynomials, e.g., B-spline, Bezier). Used for smooth animation, camera paths.

      • Translation: Moving an object from one position to another. In VR, often part of transformation matrices (translation, rotation, scale).

    • Collision Detection in Generic VR System: Determining when virtual objects intersect. Methods:

      • Bounding Volume: Simple shapes (sphere, AABB) around objects for quick checks.

      • Spatial Partitioning: Divide space (octrees, BSP trees) to reduce checks.

      • Exact Detection: Polygon-level checks (GJK algorithm) after coarse detection.

    • Real-Time Computer Graphics Fundamentals: Rendering pipeline (application, geometry, rasterization). Key techniques: Hidden surface removal (Z-buffer), Shading (Gouraud, Phong), Texture Mapping, Level of Detail (LOD). Goal: 60+ FPS for immersion.

    • Radiosity Technique: Global illumination method simulating diffuse light inter-reflection between surfaces. Based on energy conservation. Computes form factors (fraction of light leaving one surface that reaches another). Solves system of linear equations. Produces realistic soft shadows and color bleeding. Computationally intensive.

    • Acoustic Hardware in VR Systems: 3D audio systems using Head-Related Transfer Functions (HRTFs) to simulate sound direction and distance. Requires headphones and tracking to update sound source position relative to user's head.

    • Applications in Digital Entertainment: Immersive gaming, virtual concerts/movies, interactive storytelling, training simulators (flight, medical).

  • Augmented Reality (AR):

    • AR Methods & Techniques:

      • Marker-based: Uses fiducial markers (QR codes) for tracking. Simple, robust.

      • Marker-less (See below): Uses natural features.

      • Projection-based: Projects digital content onto real surfaces.

      • Superimposition-based: Replaces part of real view with virtual (e.g., IKEA app).

    • Marker-less Tracking in AR: Tracks device position/orientation without predefined markers. Uses:

      • SLAM (Simultaneous Localization and Mapping): Builds map of unknown environment while tracking location. Core for mobile AR.

      • Feature Detection & Matching: Detects natural features (corners, edges) in camera frames and matches them across frames.

      • Inertial Sensors: Accelerometer, gyroscope for motion tracking.

      • Visual-Inertial Odometry (VIO): Combines camera and IMU data for robust tracking.

  • Comparison between AR and VR:

    | Feature | Augmented Reality (AR) | Virtual Reality (VR) | | :--- | :--- | :--- | | Environment | Real world + virtual overlay | Fully immersive virtual world | | Immersion | Low to Medium | High | | Hardware | Smartphones, tablets, AR glasses (HoloLens) | HMD, often with controllers | | User Presence | In real world | In virtual world | | Primary Use | Information overlay, assistance | Entertainment, simulation, training |

  • VR Toolkits and VRML:

    • Features of VR Toolkits: Provide libraries for 3D rendering, input handling, physics, networking. Examples: Unity (C#), Unreal Engine (C++/Blueprint), OpenVR/SteamVR API. Offer scene editors, asset stores, cross-platform deployment.

    • VRML (Virtual Reality Modeling Language): File format standard for representing 3D interactive vector graphics, especially for the web. Text-based .wrl files describing scenes (nodes: Shape, Transform, Group), geometry (IndexedFaceSet), appearance (Material, Texture), and interaction (TouchSensor, Script). Largely superseded by X3D and game engines.

  • Simulation Types in VR/AR:

    • Behavior-based Simulation: Focuses on AI-driven behaviors and interactions (e.g., NPC decision-making, crowd simulation). Uses rules, state machines, or AI techniques (reinforcement learning).

    • Physical-based Simulation: Simulates real-world physics (rigid body dynamics, soft body deformation, fluid dynamics, cloth). Uses physics engines (NVIDIA PhysX, Bullet). Essential for realism in training, gaming.

[!TIP] Exam Focus: For VR/AR, know the core difference (overlay vs. immersion). For geometric modeling, distinguish wireframe/surface/solid. For radiosity, understand it's for global illumination (indirect light). For queuing theory, know Kendall notation and key metrics. For simulation steps, memorize the 6-step flow.

VII. CLOUD DEPLOYMENT MODELS (SPECIAL FOCUS)

  • Community Cloud Model:

    • Definition: Cloud infrastructure shared by several organizations from a specific community with common concerns (e.g., security requirements, compliance needs, mission, policy). It may be managed by the organizations or a third party and may exist on-premises or off-premises.

    • Characteristics:

      • Shared by a Specific Community: Not the general public (like public cloud) nor a single org (like private).

      • Common Concerns: Addresses shared needs like government regulations (FedRAMP for US govt agencies), healthcare (HIPAA), or research institutions.

      • Cost Sharing: Infrastructure cost is distributed among member organizations, making it more cost-effective than a private cloud for each.

      • Control & Customization: Offers more control and customization than public cloud, but less than a purely private cloud.

      • Governance: Often governed by a consortium or joint committee of member organizations.

  • Differences from Public Cloud Model:

    | Feature | Community Cloud | Public Cloud | | :--- | :--- | :--- | | Tenants | Specific, limited community (e.g., universities, government agencies) | General public / any organization | | Primary Goal | Serve shared community needs & compliance | Economies of scale, broad market | | Cost Model | Shared fixed/variable costs among members | Pure pay-as-you-go per tenant | | Control/Customization | Higher (community-driven policies) | Lower (standardized services) | | Security/Compliance | Tailored to community's specific regulatory framework | Generic, though often with compliance certifications | | Ownership/Management | Community orgs or dedicated third-party | Cloud provider (AWS, Azure, GCP) |

VIII. CLOUD COMPUTING PLATFORMS (OVERVIEW)

  • Major Components & Features:

    • Compute: Virtual machines (IaaS), containers, serverless functions (FaaS).

    • Storage: Object, block, file storage services.

    • Database: Managed relational (RDS) and NoSQL (Cosmos DB, DynamoDB) services.

    • Networking: Virtual networks, load balancers, CDNs, VPN gateways.

    • Management & DevOps: Monitoring, auto-scaling, configuration management, CI/CD tools.

    • Security & Identity: IAM, encryption key management, security groups, DDoS protection.

    • AI/ML Services: Pre-trained models, custom training platforms.

  • Comparative Analysis of Key Platforms:

    | Feature | Google App Engine (GAE) | Microsoft Azure | Eucalyptus | | :--- | :--- | :--- | :--- | | Primary Model | PaaS (also IaaS via Compute Engine) | Full stack (IaaS, PaaS, SaaS) | IaaS (AWS-compatible) | | Core Strength | Scalable web apps, data analytics (BigQuery) | Enterprise integration (.NET, Active Directory), hybrid cloud | Building AWS-compatible private/hybrid clouds | | Compute | App Engine (sandboxed), Compute Engine (VMs) | Azure VMs, App Service, Functions | Instances (AWS EC2 API compatible) | | Storage | Cloud Storage (object), Cloud SQL, Datastore | Blob Storage, Disk Storage, Cosmos DB | Walrus (S3-compatible), iSCSI/EBs block | | Key Differentiator | Deep integration with Google services, automatic scaling | Strong enterprise ecosystem, hybrid (Azure Stack) | Open-source, AWS API compatibility for private cloud | | Virtualization | Google's proprietary container-based | Microsoft Hyper-V | KVM (Linux), Xen (historically) | | Use Case | Cloud-native web/mobile backends, data pipelines | Enterprise lift-and-shift, .NET apps, hybrid | Organizations needing AWS compatibility on-premises |

[!TIP] Exam Focus: For platform comparison, focus on their primary positioning (PaaS vs IaaS), key differentiators (GAE's auto-scaling, Azure's enterprise/hybrid, Eucalyptus's AWS compatibility), and core compute/storage services. Know Eucalyptus is for building private clouds, not a public cloud service itself.

Go to where you left off?

Quick Add to Notes

Save questions, your own notes and screenshots into notes filed by unit. It takes a free account.

Create free account

Have an account? Log in