Cloud Computing: Definition and User Experience

Cloud Computing Definition

• Cloud computing uses internet-based computing where shared resources, software, and information are provided to computers and similar devices on demand
• Dynamically scalable and virtualized resources are provided as a service over the internet
• Gartner defines it as a style of computing that provides scalable and elastic IT capabilities as a service to external customers using internet technologies
• Users do not need expertise or control over the technology making this all possible

The New User Experience of Cloud Computing

• Cloud computing is predicted to transform information technology industry and profoundly change how people work and companies operate
• Users are shifting in how they consume information
• New enterprise applications are expected to follow a consumption model like the "Apple store"

Cloud Computing as a User Experience, Business Model, and Infrastructure Management Methodology

• Cloud computing is an emerging style where applications, data and resources are provided as services to users over the network
• Includes managing large numbers of highly virtualized resources from a management perspective and can be automatically aggregated to deliver services

NIST Definition of Cloud Computing

• Cloud computing offers convenient, on-demand network access to a shared pool of computing resources like networks, servers, storage, applications and services
• Resources rapidly provisioned
• Released with minimal management effort
• Cloud model promotes availability and includes 5 essential characteristics, 3 service models and 4 deployment models, as defined by NIST
• The 5 key characteristics include:
  • On-demand self service
  • Broad network access
  • Resource pooling
  • Rapid elasticity
  • Measured service

On-Demand Self-Service in Detail

• A consumer can unilaterally provision computing capabilities like server time and network storage, without human interaction with a cloud provider

Broad Network Access in Detail

• Capabilities are available over the network and accessed through standard mechanisms that promote use by heterogeneous thin or thick client platforms (e.g., mobile phones, laptops, PDAs)

Resource Pooling in Detail

• The provider's computing resources are pooled using a multi-tenant model where physical and virtual resources are dynamically assigned and reassigned

Rapid Elasticity in Detail

• Capabilities can be rapidly and elastically provisioned to quickly scale up and released appropriately
• Capabilities available for provisioning appear to be unlimited and can be purchased in any quantity at any time

Measured Service in Detail

• Cloud systems automatically control and optimize resource usage using a metering capability at some level of abstraction
• Resource usage is monitored, controlled, and reported, providing transparency for providers and consumers

Overview of NIST Model

• The NIST Model consists of a base of essential characteristics which is composed of Broad Network Access, Rapid Elasticity, Measured Service and On-Demand Self-Service
• Essential components are then used to provide Software (SaaS), Platform (PaaS) and and Infrastructure (IaaS) as a Service
• A final overlay allows cloud use publicly, privately, in a hybrid form or within a community

Takeaway Definition of Cloud

• Cloud is a type of dynamically-provisioned distributed system of interconnected and virtualized computers presented as a single computing resource
• Dynamic provisioning is based on service-level agreements established through negotiation between providers and consumers

Cloud Computing Characteristics

• Cloud computing is a computing paradigm and a thinking change in the IT industry without any specific technology
• Central ideas include utility computing, service-oriented architecture (SOA) and service level agreements (SLA)
• Uses enabling techniques such as hardware virtualization, parallelized/distributed computing, service/autonomic-orientated computing
• Includes properties and characteristics such as scalability/elasticity, availability/reliability, manageability/interoperability, accessibility/portability and performance/optimization

User Perspective

• Users do not care how cloud works
• They concern themselves with what they can get, the quality of service
• Users are not concerned with owning infrastructure
• Users want to pay to use the service as they need it

Utility Computing

• Computing resources and infrastructure management are made available to the customer as needed
• Charges are applied for specific resource usage rather than at a flat rate
• Utility Model seeks to maximize use of resources and/or minimize associated costs

What is Service?

• A service is what you connect together using Web Services
• A service is an endpoint to a connection
• A service should be well-defined and self-contained
• It should not depend on the context or state of other services

What are Web Services?

• A set of self describing and stateless modules that perform discrete units of work and is available over the network
• Providers offer API's to developers to exploit functionality over the Internet, rather than delivering full-blown applications
• Use Web Services Description Language (WSDL), expressed in XML, four types of operations :
  • One-way: messages sent without a reply required
  • Request and response: sending and replying messages
  • Solicit response: a request for a response
  • Notification: messages sent to multiple receivers

Service Oriented Architecture (SOA)

• Collection of services communicating with each other
• Contains flexible design principles used during system development phases and integration
• Provides a loosely-integrated suite of services to use in multiple business domains
• Implemented by modern REST model

Quality of Service (QoS)

• A set of technologies for managing network traffic cost effectively
• Enhances user experiences for home and enterprise environments
• Terms used in context of customer care evaluations (direct impact on the customer) or technological evaluations (efficient operating systems)
• The goal is to adjust procedures or adapt software to achieve the desired effect while improving use of resources

Cloud-Based SLA

• Consist of application end users and a cloud consumer who interfaces with a cloud-based business. A cloud infrastructure is provided as a service
• CA-SLA involves the business owner and cloud-based application
• CI-SLA involves the resource level and is an interface between the cloud provider and cloud resources

The Service Level Agreement

• A contract between a network service provider and a customer specifying what services the providers furnish
• The agreement is usually outlined in QoS, or measurable terms
• Includes aspects of performance metrics, problem management, penalties for non-performance and documented security capabilities. It can include Uptime/Downtime ratio, response time and system throughput

Scalability and Elasticity Distinctions

• Scalability is the ability to handle growing amounts of work and also to be readily enlarged
• Elasticity is the ability to enforce adaptive actions to enable scalability

How to Achieve Scalability and Elasticity

• Dynamic provisioning and Multi-tenant design

Horizontal vs Vertical scaling

• (a) The horizontal scaling will scale out or scale in
• (b) The vertical scaling will scale up and scale down

Horizontal Scaling

• Involves adding more computing resources (e.g., servers)
• Is a reliable, fully automated fail-over scenario
• Has a growing management complexity

Vertical Scaling

• Involves adding more powerful computing resources, but the bigger servers are a single point of failure
• Human intervention is required
• Reasonable management overhead

Dynamic Provisioning

• Explains a complex networked computing environment where server computing resources are provisioned or deployed from an administrative console/application
• i.e. Amazon auto-scaling groups

Problems with Traditional Computing Model

• System utilization is underestimated, which results in under provision
• System utilization is overestimated, which results in low utilization and higher costs

Dynamic Provisioning Recommendations

• Provision cloud resources dynamically based on seasonality, demand variations, and extraordinary events

Virtualization

• Divides hardware resources for use by machines
• A virtual machine monitor oversees running VMs
• Goal is performance isolation and security

4 Machine Interface Levels

• Unprivileged instructions are available to any program. Privileged instructions are only available to the operating system, and in turn give way to system calls
• API is a system call interface to hide implementation complexity

Types of Virtual Machines

• (a) Process Virtual Machine: multiple instances of application and corresponding runtime
• (b) Virtual Machine Monitor: allows multiple instances of apps/OS combinations

Multi-Tenant Design

• Software architecture principle where a single software runs on a server serving multiple clients' organizations
• The software virtually separated their data and configuration, allowing each client organization works with a customized application instance
• Client oriented requirements include providing customization and having a quality of service with expected adequate security levels and robustness

Availability and Reliability

• Availability is the proportion of time a system is functioning (expressed in percentage; 99.999% means "Five Nines" system is statistically available)
• Reliability is the ability of a system/component to perform its functions under stated conditions for a specified period of time
• To achieve such properties, solutions may contain fault tolerance, system resilience, and reliable system security

Fault Tolerant Systems

• Fault-tolerance enables continued operation properly in event of failure of components
• The operating quality decreases proportionally to the severity of the failure
• Four key traits are no single point of failure, fault detection and isolation of failing component, prevention of fault propagation and availability of reversion codes

Security Risk Events

• Requires building of resilient system
• Example question is can the system be restored if a key service is lost in database corruption?

Disaster Recovery

• A process/policy/procedure is needed to prepare for any continuation of technology infrastructure
• Can backup and recovery on premise to an off-site backup system

To prepare for disaster recovery

• Create local mirror systems and have uninterruptible power
• Implement surge protector and replicate data to an off-site location

System Resiliency

• Relates to providing/maintaining an acceptable level of normal service despite faults or challenges and is the ability to return to normal after trouble arises

Security Issue in Cloud Computing

• Cloud security is an evolving sub-domain of computer/network/information security and aims to protect data/applications/infrastructure related to cloud computing
• Is important to have well defended/protected data/applications
• User's data from one customer should be segregated from the data of other customers
• Enterprises should have their own system to control information access
• Provide a guarantee the application and underlying infrastructure is secure
• Data needs to be masked and restrict access unauthorized users

Manageability and Interoperability

• Manageability refers to enterprise-wide cloud systems administration
• Interoperability refers system's/product's interfaces are understood to work with other products/systems

To achieve Manageability and Interoperability

• Implement state monitoring, with built-in automation so IT staff can perform system control

System Monitoring and the Cloud

• Involves a system to collect and store state data in a distributed system and monitor system performance at all levels (hardware, resources, networks, etc)
• Is important to have a billing system

Billing System for Cloud

• In a cloud system where users pay as they go, the provider needs to determine a service usage price and have usage records
• System is monitored and then will have to be integrated into the user's banking account

Load Balancing

• Technique balances workload to get optimal resource utilization as well as minimize overload/maximize throughput
• Can involve computers, hard drives, CPUs or network links
• Load balancing may be implemented for performance and resource utilization as well as improve the efficiency of energy needs

Job Scheduling

• Software that directs background executions
• Historical name commonly known for a batch-processing-is for scheduled jobs that involve computation, workloads, and dependency

Cloud Orchestrators

• Examples include open source software packages such as Chef, Juju, Cloudify or Apache Brooklyn

IaaS Service Model

• Delivers hardware networking and virtualization services through the cloud

PaaS Service Model

• Provides tools for developers to build and host in the cloud

SaaS Service Model

• Delivers software over the cloud

Overlook of Service Models

• The customer generally manages software in IaaS, and to a less extent, the resources in PaaS
• In SaaS model all functions are managed by the vendor

Infrastructure as a Service

• Provides fundamental computing resources such as processing, network, and storage
• Has examples such as Amazon EC2, Eucalyputs or Openstack
• The costumer deploys software or services in the cloud
• Does not manage the underlying cloud infrastructure

Virtualization Layer Enables and Supports:

• Manageability/Interoperability, HA, and Scalability

Virtual Infrastructure Manager (VIM)

• A type of software used by IaaS providers for cloud infrastructure, managing and provisioning across physical hardware
• Resembles a traditional operating system for a single computer, although handles resources across multiple systems and presents the uniform result to the users

Virtual Machine Monitoring

• Involves the ability to monitor and control virtual resource activity/system operation as well as provide basic resources like virtual machines/storage/networks, etc.
• Several monitoring metrics are used involving data about VM state, storage state or network activity

AWS Auto Scaling

• AWS is a web service where users can specify how resources change in response to changing coditions

Key Concepts of Availability Zones in AWS

• The Availability Zone is located with an AWS Region
• Each availability zone is an independent center with its own unique power grid and network connections
• This is to protect various failures between the zones
• Zones are also have low latency connections

Key Concepts of Images vs Instances

• Amazon Machine Image (AMI): The blueprint for launching virtual servers in AWS with operating and application layering. It has a unique ID and can be customized or shared or even sold
• Elastic Compute Cloud (EC2) Instance: Represents running copy of an AMI. Any number of copies of the same AMI can be launched. Represents a running copy of AMI. Allows use of Elastic IP Addresses or static IP addresses to instances

Key Concepts of Elastic Block Store (EBS) Volume

• EBS volumes addressable disk volumes that can be attached to running instances in the same AVA zone. EBS volumes can be formatted, mounted, and used as local drives
• The volume can persist after the system is turned down

Key Concepts of the Security Group

• Allowable network settings for inbound connections; can be regulated by multiple groups including list of protocols, ports and IP address ranges
• A group can have multiple instances

Operation of Auto Scaling

• Ensures EC2 fleet are performing optimally, will terminate an instance and launch another. If any instance terminates unexpectedly, it will replace with new replacement automatically

AWS Auto Scaling Details

• It enables automated launching/termination of EC2 instances based on user-defined policies, schedules or health checks

• It is useful for fleet of managing fleet of EC2 instances handles a presents load as well as to guarantee running EC2 instances

• Can seamlessly increase and decrease instances

• Is targeted towards applications that experience hourly, daily or weekly usage variability

Health Check

• An ongoing call to check each instance and terminate launch a new one that takes place when health check fails

Launch Configuration

• Includes parameters needed to create an EC2 instance. New or updated launch configuration parameters launch updated parameters

Auto Scaling Function

• Functions to cover volume of traffic, including EC2 and URL's/web addresses. Instances are launched using a set of min/max of the service

Auto Scaling Is A Group

• Group shares similar characteristics for the purpose of scaling and management that can be contained from one or more EC2 availability zones

Details of the Launch Configuration

• Every group has a launch configuration and Launch configurations enable to describe each instance that Auto Scaling
• Is controlled by a description

Triggers in AWS scaling

These indicate launching or removing of data and activated by watching CPU utilization

• A Trigger is a mechanism that tell the system when to increase or decrease the number of instances

• When activated, the trigger launches a long-running process called Scaling Activity supports

Concepts for a Trigger

• To achieve, the Trigger is includes a cloud watch system and autoscaling policy of description
• In most cases, two triggers are needed with alarms for scaling up and scaling down

Alarms

An object that watches a single metric

Policies in AWS

• Sets of instructions and alarms that triggers cloud watch

AWS Scaling Policies

Policies long-running processes that change a group based on new/replaced instances

EBS Auto Tracking

• If CPU increases past 80%, use cloudwatch. Autoscaling detects what is to do. Create alarm for 40%

Understanding the Service Models

IaaS Model

• Infrastructure as a Service provides fundamental computing resources such as processing, network, and storage
• Offers examples such as Amazon EC2, Eucalyputs or Openstack
• Costumer deploys software or services, but does not manage the underlying cloud infrastructure (virtualization later enables/supports better management, HA, and scalability)

PaaS Model

• Platform as a Service Provides tools for developers to build and host apps
• The integration supports development, running and management, is the most important consideration

SaaS Model

• Software as a Service is applications/services delivered over a cloud
• Provides General, Business, Scientific, Government Applications and web portal
• Employs web 2.0 to maximize communication potential from its web based services and is easily accessible over the internet
• These techniques enables you to rent and customize services

Private Cloud Model

• Cloud infrastructure is operated solely for an organization, but can be managed in-house or as a third party
• Is also commonly known as internal or on-premise cloud. Characterized as tailored policies/end to end with dedicated/tailored policies/In-house (or Heterogeneous)

Characteristics in Cloud vs Private

• Public cloud's characteristics are homogeneous, multi-tenant operational expenditure and large-economy
• Private Cloud's characteristics are heterogeneous, dedicated and custom with end-to-end control

Community Cloud Model

• Cloud infrastructure shared by several organizations with shared concerns. Is known is a collaborative and cost effective approach

Hybrid Cloud Model

• Cloud infrastructure is a combination of two or more bound clouds (private, comm or pub)
• Is valuable for data load-balancing with portable applications

Cloud Market Share Over The Years

• Amazon is the market share leader, followed by Microsoft, Google, IBM, Oracle and Salesforce
• The role of is extremely important

Cloud computing delivers scalable IT resources as a service over the internet. Users consume information in new ways, and enterprise applications are adopting models like the 'Apple store'. This transforms IT and how businesses operate.