When cloud provider provisions fundamental computing resources such as networks, processing, and storage, it is understood that the consumer has hardly any control on underlying physical infrastructure of cloud. Consumer of these resources is able to control deployed applications, operating systems, storage and to networking components to some extent.  

Cloud Computing Parameters

There has been a significant confusion regarding a proper definition of Infrastructure as a Service. This is because of cloud service providers are offering a large variety of features and services that are purported to be focused at Infrastructure as a Service. There has to be an in depth understanding of the special characteristics that must be offered by cloud provider in order to qualify the offered service as ‘Infrastructure as a Service’.

Thanks to evolving technologies, NIST (National Institute of Standards and Technology) has provided a comprehensive definition of cloud computing that comprises of:

•    Five significant characteristics
•    Three basic service models
•    Four important models of deployment

Essential Characteristics of Cloud Computing

Self Service As Per Demand - There must be an arrangement that facilitates consumers to unilaterally and individually access important computing capabilities including storage, compute power, and network connectivity. This should be provisioned automatically with no human intervention.

Broad Access To Network- Ability to access network via standard mechanisms should be focused at using variety of thick and thin platforms.

Pooling of Computing Resources - In order to provide benefits of a multi tenant model to more consumers, all resources of the provider including physical as well as virtual resources are pooled together to dynamically assign or reassign these in response to demand from consumers. The resources can comprise of virtual machines, processing power, storage, memory, and network bandwidth.  

It needs to be understood that although consumers of cloud services are not able to control precise location of resources, the exact location at higher level of abstraction including a data center, region, or country may be specified by providers. 

Essentially Cloud Services are Governed by Location Independence.  

Rapid and Elastic - Rapid as well as elastic provisioning of capabilities support rapid scaling providing a sense of unlimited resources to the consumer. These capabilities can be purchased irrespective of quantities and time.
 
Metered Service - Resource consumption of cloud hosting services is controlled and optimized automatically with help of measuring functionality at an appropriate abstraction level according to the type of service. Metering of service may cover active user accounts, compute, bandwidth, or storage and be capable of monitoring, reporting, and controlling resource consumption.  There must be transparency of service utilization for consumer as well as provider. 

Cloud Computing Service Models

Cloud SaaS - Software as a Service is the capability accorded to the consumer for accessing and utilizing applications running on cloud infrastructure of provider. These applications can be accessed via web browser from different client devices. The underlying cloud infrastructure components such as servers, network, and operating systems, are not accessible to the consumers. 

Cloud PaaS - In Platform as a Service, consumers are allowed to deploy applications that are created or acquired by them on provider’s cloud infrastructure with help of tools and programming languages supported by service providers. Consumers are able to control operating systems and storage of deployed applications. However they will not be able to control underlying cloud infrastructure. 

Cloud Deployment Models

Private Cloud Deployment Model - This model is entirely operated by the organization for its own purpose. The management may be assigned to the third part or retained with the organization itself either on-site or off-site. 

Community Cloud - This arrangement is designed to serve interests of an exclusive community consisting of multiple enterprises that share common purpose such as policy, mission, compliance needs, or security concerns. 

Public Cloud Deployment Model - This deployment model is offered to public at large or a big group of industries. Public cloud is owned by any provider organization that is engaged in selling cloud solutions. 

Hybrid Cloud Model - The hybrid cloud model is created by combining two or more clouds that retain their individual entities in spite of being clubbed together to enable portability of applications including load balancing and cloud bursting. Hybrid cloud is characterized by permutations of private cloud, public cloud and community cloud.

Attributes of Reference Architecture

Ideal reference architecture needs to be in accordance with NIST definitions. This should cover all patterns, concepts, and core principles across the entire reference architecture. Once the reference architecture is established as per NIST definition, then it is essential to design service delivery guidance for directing the business. 

New Paradigms of IT Delivery

Thanks to cloud computing, businesses can adopt entirely new approaches for sourcing and controlling IT services. Cloud is capable of delivering a well defined set of solutions that are viewed by customers to have unlimited capacity, greater cost efficiency, and remarkable agility. In order to make customers appreciate these attributes, IT needs to shift from server oriented mindset to service oriented approach. The approach moves away from silos for application deployment to standardized platforms by accepting mutually acceptable SLAs. This calls for a broad strategy of hybrid cloud that accommodates a mix of different models of cloud to achieve desired objectives. 

SaaS 
Consumers of SaaS can be offered applications or business process including CRM, email, or collaboration. It abstracts every function associated with IT operations and infrastructure. The operational responsibilities and related risks are shouldered by the provider of SaaS which is a highly cost efficient and intuitive option. 

PaaS
This is associated with delivery of application execution services including integration, storage, or runtime. Consumers are able to achieve an agile and efficient ability for running scalable applications in a defined and economical way. In PaaS model the operational risks are equally shared between consumer and provider.
 
IaaS
The IaaS model abstracts hardware such as storage, servers, or networking infrastructure to provide a pool of different capabilities including connectivity, storage, and computing. These capabilities are delivered at measured costs as services. The ultimate aim of IaaS is to develop standard, virtualized, yet flexible operating environment to present a robust foundation for delivering SaaS and PaaS. 

Interesting Topic:

Advantages of Cloud Computing

(PaaS​) Platform as a Service