Cloud Computing: Principles and Technology

Cloud Computing: Principles and Technology (4IT482)

• Key course topic: Cloud Computing
• Course date: January 2025
• Instructor: Dr. George Feuerlicht
• Institution: Prague University of Economics

Introduction

• IT trends
• Historical perspective of IT
• Technology and business drivers
• Definition of Cloud Computing
• Course objectives and topics

Learning Objectives

• Understanding cloud computing motivations
• Understanding business and technological drivers
• Appreciating the evolution of cloud computing
• Recognizing the benefits of cloud computing

IT trends and predictions

• 74% of executives expect AI to change business processes within 5 years.
• Al-leveraging companies are expected to outperform peers by 25% in profitability by 2025.
• Generative AI market will explode from $40 billion in 2022 to $1.3 trillion by 2032.
• Generative AI could raise global GDP by 7% and save workers up to 70% of time, according to McKinsey.

Historical Perspective

• 1960s: Centralized mainframe computing
• 1970s-80s: In-house development
• 1990s: ERP applications
• 2000s: Outsourcing and ASP models
• 2007: SOA/SaaS
• 2010: Cloud computing

ASP (Application Service Provider) Model

• Precursor to cloud computing (early 2000s)
• Application services over a network
• Unsuitable architecture
• Poor scalability
• Connectivity issues

Utility Computing

• Information technology's strategic importance has diminished
• Companies are shifting from owning IT assets to purchasing IT as a service.

Commoditization of IT

• The internet has accelerated the commoditization of IT.
• Companies purchase web-based services similar to purchasing electricity or telecommunications.

Google Data Centre Containers

• Extension of data centers with self-contained units
• Includes machines, air conditioning, and power supply

20th-Century Paradigm

• Software vendors license software
• Customers install, customize, and maintain software
• Under-utilization of hardware resources
• Costly, inefficient, and unsustainable

On-premises Costs

• Capital expenses
• Technical personnel costs
• Administrative staff costs
• Frequent project failures
• Clients need to focus on core business, not IT

21st-Century Paradigm

• Fast and reliable connectivity
• Scalable computer infrastructure
• Economies of scale in data centers
• Services delivered by software vendors
• On-premises deployment less relevant

Technology Drivers

• Moore's Law: Increasing transistor counts on integrated circuits
• Increased processing power
• Increased storage capacity
• Increased network bandwidth
• Reduced cost and size
• Advanced processor architectures
• Network effect and economies of scale
• Reduced cost of shared services as user numbers increase

Fast Connectivity

• Ultra-fast networks like 5G and Wi-Fi 6E
• Lower latency supports cloud-based applications
• Supports near real-time applications

Network Effect - Economies of Scale

• Product/service cost reduces as consumer numbers increase
• Product/service value increases with user numbers

Business Drivers

• Reduction of cost of entry (start-up costs)
• Reduction of TCO (Total Cost of Ownership)
• Minimizing hiring of technical staff
• Avoiding utility costs
• Scalability (horizontal and vertical)
• Flexibility (adaptability)
• Global reach (multiple regional deployment)

What is Cloud Computing?

• Delivery of virtualized IT resources over the internet
• Scalable and secure computing services on demand.
• Pay-per-use basis
• Infrastructure services (IaaS), Platform services (PaaS), and Software as a Service (SaaS)

Cloud Computing

• Elastic IT services delivered on demand
• Benefits include agility and ease of use.
• Cost-predictability
• Global deployment, and focus on core business

20th century paradigm characteristics

• Software vendors license software, and customers install and maintain.
• Under-utilization of hardware is a common problem.
• IT skills are intensely demanded within organizations.

Cloud computing

• IT service delivered on demand via the internet
• Economies of scale
• Flexibility in scaling services up or down

Distributed Computing Concepts and Techniques

• Evolution of distributed computing
• Client/server Architecture
• Remote Procedure Calls (RPCs)
• Message Queuing
• Application Programming Interfaces (APIs)
• Transactions and Consistency
• Data Replication
• Service Oriented Architecture (SOA)
• Microservices Architecture

Enterprise Computing Objectives

• Functionality as per user requirements
• Maintainability (flexibility and reuse)
• Reliability (consistency, fault tolerance, recovery)
• Elasticity (up and down scalability)
• Security (authentication, authorization, physical security)
• Cost minimization (ROI)
• Rapidly evolving technology

IT Architecture

• Decisions about implementing application layers, such as what devices/platforms they should run on, how they communicate with each other and how to maximize performance while minimizing the cost.
• These factors change as IT evolves.

Evolution of Distributed Computing

• Centralized mainframe computing (1960s)
• Client/server Computing (1980s)
• Distributed Objects and Components (1990s)
• Service Oriented Computing (2000s)
• Cloud & Microservices Architecture (2010-present)

IBM 1800 Computer

• 64KB memory
• 1-5 MB disks
• Punch cards
• Magnetic tape
• Fortron/Cobol

2-Tier Client/Server Architecture

• PC Workstations
• Database Server

3-Tier Client/Server Architecture

• Client
• Middleware
• Database Server

Client/Server Communications

• Synchronous vs. Asynchronous communication
• Message vs. Procedure Calls (RPCs)

Remote Procedure Calls (RPCs)

• Basic client/server communication mechanism.
• Supported at programming, database, and operating system levels.

Message Queuing

• Communication model for highly distributed applications.
• Suitable for heterogeneous and autonomous applications.
• Resilient to network and application failures.

Transactions

• Centralized and distributed DBMS
• TP monitor middleware
• CORBA (Object Transaction Service/OTS)
• EJB (Java Transaction Service/JTS)
• Web services
• NoSQL databases

Full Consistency Model

• Synchronous tightly coupled transactions
• Consistent state reached after every transaction
• All operations (sub-transactions) complete in a context of a single atomic transaction
• Resources required to complete the transaction
• Failure recovery required

ACID Transactions

• Atomicity (all operations or none)
• Consistency (data consistent after completion)
• Isolation (partial results not shown until transaction completes)
• Durability (committed transactions cannot be undone)

Commit and Rollback

• Rollback recovery takes place when a failure occurs during transaction execution.
• This restores the database to its original state.
• Roll-forward recovery involves re-applying committed transactions.
• A database backup is required for this recovery.

Distributed Transactions

• Coordinator (C) and other participants (P)
• Two phases: establishing if they are ready to commit and then committing.
• Participants record events in local logs, before sending messages.

2PC Suitability

• Synchronous protocol resilient to failures.
• Complexity and performance issues
• Blocking
• Scalability issues (all participants must be available)

Eventual Consistency Model

• Asynchronous operation
• Independent sub-transactions
• Loose consistency in that eventually all data is consistent
• Good for distributed systems with high latency and low reliability

Data Replication

• Multiple copies of data objects
• Replication transparency for managed redundancy
• Replication can improve performance and availability
• Used to synchronize data among multiple databases or storage systems

Synchronous Replication

• Tight consistency model
• All copies are always immediately up-to-date, updated in a single transaction
• Suitable for reliable, fast networks
• Needed for applications that cannot tolerate asynchrony

Asynchronous Replication

• Loose consistency model
• Updates are deferred, inconsistencies may occur
• Suitable for unreliable, intermittently connected networks
• Tolerates some degree of asynchrony

Master-Slave Replication

• Table snapshots
• Regularly updated in intervals
• Full or incremental refresh
• Can be performed during off-peak periods

Multi-Master Replication

• All sites are equal
• Updates targeted at any copy
• Conflicts possible with latency intervals

Conflict Resolution

• System detects conflicts.
• Application defines the resolution algorithm.
• Standard resolution algorithm: timestamp.
• Commutative resolution of additive updates.

Application Programming Interfaces (APIs)

• Gain independence of the underlying platforms.
• Portability across different platforms.
• Standardizing APIs avoids point-to-point integration solutions.
• Extensive use in web and cloud applications

Service-Oriented Architecture (SOA)

• Set of architectural concepts, principles, design patterns, and technologies.
• For implementation of service-oriented applications.
• Basic SOA abstraction = services
• Services are autonomous and loosely coupled.
• Standards for machine-to-machine communication (SOAP or REST)

SOA Motivations

• Requirements to support inter-enterprise business processes.
• Outsourcing individual business functions to external providers
• Need for a more flexible and responsive intra-enterprise computing architecture
• Supporting use of different languages, in conjunction with interfaces such as XML and JSON

Properties of Services

• Functional Properties: WSDL, Service Methods, Protocols.
• Non-Functional Properties (QoS): Security, Availability, Response time, Price, etc.

Web Services Standards

• Service Discovery (UDDI)
• Orchestration/Composition (WS-BPEL)
• Service Description (WSDL)
• Messaging (SOAP/HTTP/XML)
• Security (WS-Security)
• Transaction management (WS-Transactions)

SOAP

• Standard protocol for communication between services.
• Mechanism for error handling.
• Extensibility mechanism.
• Conventions for representing data structures in XML
• Supports remote procedure calls (RPCS) & XML document interchange

REST (Representational State Transfer)

• Architectural style and design of network-based software.
• Stateless operation; each request contains all the necessary information.
• Named resources using URLs.
• Uniform API: all resources accessed through a generic interface (HTTP GET, POST, PUT, DELETE)
• Uses web infrastructure like HTTP, but not SOAP

Microservices Architecture

• Application development as loosely-coupled, independently deployable services, and well-defined protocols
• Each independent service typically implements a single(cohesive)function.
• Implemented using containers (often Docker)
• Light-weight messaging protocols to facilitate communication between services
• High scalability and fault tolerance using container replication

DevOps (Development and Operations)

• Combines software development and software operation.
• Automated monitoring of the SDLC process.
• Smaller development teams, shorter cycles, high deployment frequency.
• Closely linked with microservices and cloud computing

DevOps and Microservices Architecture

• Microservices are typically implemented using Docker containers, and communicate via stable APIs
• Cloud deployment and scalability improvements by containerization.
• DevOps teams react quickly to user requests for enhancements and fixes.

Scalability and Fault Tolerance Considerations

• Modern cloud applications require fault-tolerance and scalability.
• Vertical scaling improves performance by adding resources (CPU, memory, storage, etc.) to an existing computing node.
• Horizontal scaling improves performance by adding more nodes which share processing.
• Fault-tolerance is achieved by running multiple identical microservices on multiple nodes.

Horizontal Scalability

• Share-nothing architecture where each node contains all necessary resources (CPUs, RAM, storage etc).
• Increase cluster nodes to improve scalability.
• Shard data into partitions to increase storage capacity.
• Synchronous data replication enables high availability by creating 3-4 copies of every data record.

Fault Tolerance

• System capacity needs to exceed expected loads.
• Unforeseen events such as node failure and power interruptions can lead to failure.
• Redundancy is included in any critical component (e.g. backup systems).
• Recovery mechanisms are included using stand-by systems or snapshots to recover from a failure.

Summary

• Cloud computing architecture evolves to handle more powerful, fast, and less expensive hardware components.
• Cloud computing leverages large numbers of independent commodity computing units (nodes).

What is Cloud Computing? (Lecture 3)

• Cloud computing is an umbrella term with no precise definition.
• Most recent definitions focus on multitenancy and scalable resource allocation

NIST Definition

• Cloud computing is a pay-per-use model for enabling convenient, available, on-demand network access to a shared pool of configurable computing resources

Cloud Computing Pre-requisites

• Reliable and low-cost communication
• Virtualized resources that can scale
• Fault tolerance

Virtualization

• Infrastructure as software.
• Improves efficiency and agility
• Faster provisioning
• Isolates applications
• Independent of hardware

Virtual Machines

• Pre-configured images of applications / OS
• Abstraction of physical hardware
• Administered by a hypervisor
• Can be slow to deploy

Container-Based Virtualization

• Application abstraction that packages application and dependencies
• Multiple containers can share the OS kernel
• Resource use minimized as containers use less space than VMs
• Easier provisioning and recovery

Kubernetes Project

• System for automating the deployment, scaling, and management of containerized applications
• An open-source cluster manager initially created by Google for Docker containers.
• Hosted by the Cloud Native Computing Foundation (CNCF).

Kubernetes Concepts

• Environment for services, not machines
• Abstraction of the complexity of the underlying cloud infrastructure
• Auto-scaling and dynamic scaling features

Multitenant Architecture

• Tenants operate in virtual isolation from one another.
• Each tenant has its own virtual database for isolation.
• Multitenant data models enable extending the database by adding columns to tables or new database objects.

Multi-instance vs Multitenant Architecture

• Multi-instance = separate instance for each tenant, not suitable for scaling
• Multitenant = same instance for all tenants

Polymorphic Application

• Applications that adapt to different user interfaces, business logic, database schemas, and runtime behaviors in line with tenant needs.
• Application components are configured at runtime, giving each tenant the option to use different versions of application modules.

Cloud Service Models

• IaaS (Infrastructure as a Service): Management of computing resources.
• PaaS (Platform as a Service): Development & delivery of applications.
• SaaS (Software as a Service): Complete apps (emails, CRM, etc).
• Different levels of control and responsibility

AWS (IaaS)

• Amazon EC2 Compute Services - computing resources.
• Amazon S3 storage - data storage
• Amazon RDS database services - relational databases
• Amazon NoSQL database services
• Amazon VPC service - virtual networking

Microsoft Azure (IaaS)

• Virtual Machines (Windows and Linux)
• Networking (virtual networks, load balancers, VPN gateway) etc.

Google App Engine (PaaS)

• Java development and deployment platform.
• Supports Java Servlets and other standard Java frameworks.
• Language-runtime support for many languages

Salesforce.com (SaaS)

• Platform for CRM applications and infrastructure.
• Meta-based customization for apps to adapt to different tenants
• Meta-based customization to allow users to retain changes to their applications during software upgrades

Function as a Service (FaaS)

• Serverless execution model that dynamically adjusts compute resources to specific code requests.
• The cloud provider handles provisioning and maintenance of execution resources.
• The user pays only for actual execution time.

Characteristics of Serverless Computing

• Support for event-driven programming
• Fully managed stateless services
• Rapid transparent scalability
• On-demand service
• Built-in logging and monitoring
• Represents the next step in virtualization

Cloud Deployment Models

• Public cloud: Owned and managed by the provider
• Private cloud: Owned or leased by a single organization
• Hybrid cloud: Combination of public and private clouds

Government Cloud

• AWS GovCloud (US) is an isolated AWS region.
• Customers can host sensitive workloads in this region.

Benefits of Cloud Computing

• Reduced cost
• Predictability of costs (pay-as-you-go)
• Elasticity (up and down scaling)
• Possibility of short-term IT resource use
• Risk transference (provider responsible for operations and upgrades)
• Encourages innovative solutions

Cost Reduction

• Improved hardware utilization (increased efficiency)
• Savings made possible by data center locations that minimize staffing, electricity, and accommodation costs.

Cloud Computing Challenges

• Customer lock-in
• Data confidentiality and security
• Data transfer bottlenecks
• Performance unpredictability

Business Continuity and Service Availability

• Multiple data centers in different locations and network providers
• Increased scalability and elasticity
• Round-robin HTTP request distribution

Provider Liability

• No liability in case of unauthorized access or data loss

Data Confidentiality and Security

• Auditability requirements (such as Sarbanes-Oxley)
• Cloud providers must guard against data theft & denial-of-service attacks
• Data must be protected in the cloud

Data Transfer Bottlenecks

• Data transfers can be expensive
• AWS offers physical shipping of storage devices as a cheaper option

Performance Unpredictability

• Variability in disk write bandwidth
• I/O interference between virtual machines
• Unpredictability in running large batch jobs in large clusters

Identifying Opportunities for Cloud Deployment

• Good solution for unpredictable demand on compute and storage resources (on-demand provisioning)
• Suitable for startups.

Conclusions

• Most enterprise applications are delivered via cloud services
• Carefully evaluating cloud scenarios is important.
• Standardization efforts for cloud-native solutions are in progress.
• Cloud services may be affected by the industry trend towards specialized processor chips and use of Artificial Intelligence.

Amazon Web Services (AWS) Core Services

• Provisioning IT resources (compute, network, database, etc.) on-demand using pay-as-you-go model.

AWS Compute Services

• Types of Compute including EC2, AWS Lambda, container-based services such as ECS and EKS, and web applications (AWS Beanstalk).
• Selection of service based on application design, availability requirements, use case, and usage pattern.

AWS Classic Load Balancer (CLB)

• Layer 4 (Transport Layer) of the OSI model.
• Single entry point for accessing several EC2 instances and distributes traffic across availability zones (AZs).
• Provides high availability (HA), fault tolerance and increased scalability and elasticity

AWS Application Load Balancer (ALB)

• Layer 7 (Application Layer) of the OSI model.
• Enables native support for microservices and container-based architectures.
• Directs traffic based on the content of URLs.

Auto Scaling

• Auto Scaling ensures that the correct number of EC2 instances are available to meet the workload requirements.
• Minimizes costs by adjusting the number of instances as needed.
• On-demand provisioning
• AWS CloudWatch monitoring facilitates dynamic adjustment to peak load

Cost Optimization Options

• On-Demand Instances
• Spot Instances
• Reserved Instances
• Dedicated Hosts

AWS Lambda

• Serverless computing that dynamically allocates compute resources based on code requests.
• No server management or provisioning needed.

AWS Lambda Use Cases

• Image recognition.
• Automated backups, IoT applications
• Uploading to storage such as S3.
• Processing real-time streaming data.

Amazon Elastic Container Service for Kubernetes (EKS)

• Runs Kubernetes on AWS infrastructure, across multiple availability zones.
• Automatic replacement of unhealthy instances.
• Automated version upgrades
• Supports automatic scaling for performance metrics.

Horizontal Pod Autoscaler

• Uses metric server to collect resource utilization metrics.
• Creates Pod replicas on demand based on these metrics.

Machine Learning (ML) Services

• Speech Recognition (Polly)
• Image Recognition (Rekognition)
• SageMaker (building models)
• Amazon Code Whisperer (coding recommendations)
• Fraud Detector (detection of fraud)

Automating IT Operations (AWS CloudFormation)

• Creates resources such as EC2 instances, Amazon RDS databases, and Elastic Load Balancers.
• Manages resource provisioning and recovery from failures in an automated and predictable manner

AWS Billing and Free Tier

• Track costs and usage.
• Analyze cost drivers and usage trends.
• AWS Free Tier provides free usage of services for a set period of time.

Summary

• Summarizes and synthesizes the main ideas.

NoSQL Databases (Lecture 6)

• Database trends, motivations for NoSQL, types of NoSQL data stores, examples of NoSQL databases, including DynamoDB, MongoDB, and Neo4j
• Considerations of database schema, data consistency issues and potential solutions

Big Data

• Amount of data generated from various technologies, including IoT, is growing exponentially.
• Structured Data forms a small percentage of total data generated.
• Large-scale computing applications using NoSQL are necessary to manage the increase of unstructured data.

What Problems Are NoSQL Databases Addressing?

• Modern database challenges, very large data sizes, and very large user populations.
• Data complexity issues; need for reliability in a distributed architecture.

NoSQL Databases

• Key-value stores (e.g., DynamoDB).
• Column-oriented databases (e.g., Vertica)
• Document databases (e.g., MongoDB, CouchDB).
• Graph databases (e.g., Neo4j).
• Other examples include Google BigTable.

Share-nothing cluster architecture

• Horizontal scaling involving thousands of machines.
• Data duplication to increase availability.

CAP Theorem

• Consistency, Availability, Partition Tolerance.
• There is a tradeoff between these three properties in any distributed system.

Eventual Consistency

• Acceptable consistency for some application scenarios.
• Focus is on availability over consistency in the presence of network partitions, but consistency will be enforced eventually.

DynamoDB Features

• No practical limit on the number of items
• Low-latency queries (less than a millisecond)
• Scalable read/write operations
• Automatically provisions throughput
• Data indexing and partitioning by primary keys
• Query by primary keys or scan requests

DynamoDB Tunable Consistency

• Offers the options for selecting consistency levels of writes ranging from strong consistency to eventual consistency.
• W: Number of copies written synchronously.
• N: Total number of replicas.
• R: Number of replicas that must be read to achieve consistency.

DynamoDB Case Studies

• Companies like Zoom, Netflix, Disney, and Dropbox utilize DynamoDB for their operational needs.

MongoDB

• JSON-based document database that supports master-slave replication to handle high-volume read requests
• Data automatically distributed on physical nodes (shards)
• A router acts as an intermediary for requests to the relevant shard

MongoDB Logical Structures

• Database, Collections, Documents, Fields, Embedded Documents, References

MongoDB Horizontal Scaling using shards

• Distribution of data across numerous physical nodes.
• Sharding key is used for data partitioning.

Replica Sets

• Provides master-slave replication of data via synchronous updates to the master.

Replication

• Master-slave replication for propagating updates to replicas
• Replication can span different regions.

MongoDB Tunable Consistency

• Different write and read concerns can be defined and customized.

Transactions

• Operations, typically across multiple documents in a single collection.

Indexing

• MongoDB uses indices to increase query performance.

MongoDB Demo

• Practical demonstrations using querying and retrieval of data from a sample dataset

Aggregation Framework

• Tools to efficiently aggregate and group data in MongoDB.

Neo4j

• Open source graph database.
• High scalability
• Master-slave replication support.
• ACID transactions.

Property Graph Model

• Represent distinct objects as nodes.
• Relationships between objects represented as edges.
• Properties for both nodes and relationships to describe related data.

Example Graph

• Example of nodes, relations, and properties to represent an example of data relationship

Neo4j Demo

• Examples of queries for finding, updating, and manipulating specific properties in a graph of related data

Summary

• Summary of NoSQL databases based on what was covered

Machine Learning (ML) - Lecture 7

• Introduces Machine Learning concepts, models (predictive models and generative models), AWS services, and hands-on practices (exercises)

Amazon ML Services

• Overview of Amazon ML services (e.g., Rekognition, Transcribe, Translate)
• Focus on services and examples (e.g. Amazon Rekognition)

Amazon Rekognition

• Object, Scene, Activity, Facial, and Celebrity recognition and analysis.
• Video analysis - movement, activities.
• Text in images analysis

Amazon Transcribe

• Transcribes speech to text.
• Enables intelligent punctuation and grammatical formatting.
• Provides timestamps and confidence levels for extracted words.
• Offers custom vocabulary option for improved accuracy (e.g., with brand names, etc)

Amazon Comprehend

• Discovers insights and relationships in text.
• Locates named entities and keyphrases in a document.
• Identifies the language of the text and detects positive/negative sentiment in it

Amazon Polly, Translate, Lex

• Amazon Polly - Text to speech transformation
• Amazon Translate - Real-time or batch mode translation, 127 supported language pairs
• Amazon Lex - Builds voice interfaces (and provides access to Lambda functions for complex use cases)

Amazon Q/Whisperer

• Suggests code improvements based on existing code and comments.
• Highlights security issues in code.
• Integrates with multiple programming languages and IDEs.

Summary

• Overview of ML learning, including practical application, emphasis on standardization of systems, and leveraging industrialization methods

Cloud Computing Adoption (Lecture 8)

• Cloud Computing adoption Framework and AWS Adoption Framework Perspectives, stages of cloud adoption, including the challenges
• Specific AWS features and considerations of cloud implementation.

Cloud Adoption Framework (SaaS)

• Detailed cost-benefit analysis needed to maximize benefits and minimize risks of cloud computing.
• Considerations that include: utilization patterns, data sensitivity, service provider suitability, integration issues

Cloud Adoption Framework

• Evaluate business impact, including risk and organizational readiness.
• Evaluate applications and selection of appropriate providers.
• Plan cloud migration and governance structures.

Strategic Phase

• Risk management and cloud provider risk mitigation
• Security and data privacy considerations
• Business continuity plans; including disaster recovery
• Business integration, customization requirements and regulatory compliance

Tactical Phase

• Evaluation of suitable applications for cloud implementation
• Provider selection

Selecting Applications for Cloud

• Utiization patterns and demand variability/
• Data sensitivity and privacy considerations
• Integration requirements for APIs
• Customization needs
• Availability requirements
• Response-time requirements
• Elasticity requirements

Selecting Cloud Service Providers

• Provider stability
• Pricing model and costs
• Provider dependence considerations
• Standard API support
• Data lock-in considerations
• Levels of trust

Implementation Phase

• Pilot implementation to verify all the functional and non-functional requirements.
• Defining tasks and milestones to support application migration.
• IT operations adaptations to support and manage applications.
• Governance structures for defining roles/responsibilities

AWS Cloud Adoption Framework Perspectives

• Business Capabilities
• Governance: priorities, control, and risk
• Technical capabilities
• Platform: applications, infrastructure
• Security
• Operations

Business Perspective: Value Realization

• IT finance
• IT alignment with business strategies
• Benefits realization (TCO)
• Risk management (strategic, external)

People Perspective: Roles and Readiness

• Human Resources, Staff, and people managers
• Evaluating organizational structures.
• Identifying new skills and defining requirements.
• Prioritize training needs.

Governance Perspective: Prioritization and Control

• Program and Project Management
• Business Performance Measurements
• License Management

Platform Perspective: Applications and Infrastructure

• Network Provisioning, Storage Provisioning, Database Provisioning, Application Development

Security Perspective: Risk and Compliance

• Structure security controls and practices to address organizational objectives with regards to visibility, auditability, control, and agility
• Includes detective mechanisms for event monitoring, infrastructure security controls, and data protection considerations for usage

Operations Perspective: Manage and Scale

• Service Monitoring and issue and problem resolution.
• Resource Inventory Management.
• Release and Change Management
• Reporting and Analytics – compliance with KPIs (Key Performance Indicators).

Cloud Economics (AWS)

• Cost Components (compute, storage, data transfer)
• Data transfer costs within regions (free)
• Volume discounts
• Free-tier services

Total Cost of Ownership

• Cloud vs. on-premises.
• Includes capital and operational costs.

Return on Investment (ROI)

• Profit (minus total cost of ownership) divided by total cost of ownership.

Summary (Lecture 9)

• Cloud Computing Redefines aspects of enterprise computing