Computing over the cloud: Datacraft

BANGALORE, INDIA: Cloud computing and its catalysts, will impact every area and line of business (LoB) and every service model we have today.

Cloud computing can be defined as a style of computing where massively scalable IT-enabled capabilities are delivered 'as a service' to external customers using Internet technologies.

Is India Ready For Cloud Computing?

The cloud-computing model is changing the way the IT industry looks at user and vendor relationships. As service providers, vendors must partner with service providers to deliver their technologies indirectly to users.

User organisations will watch portfolios of owned technologies decline as service portfolios grow. The key activity will be to determine which cloud services will be viable, and when. Although it is unlikely that users will completely abandon on-premises models, or buy complex mission-critical processes as services through the cloud in the next few years, there will be a movement to consume services in a more cost-effective way.

The key characteristics of cloud computing include, delivery of capabilities "as a service," delivery of services in a highly scalable and elastic fashion, using Internet, web — technologies and techniques to develop and deliver the services, and designing for delivery to external customers.

Any idea who invented cloud computing?

The final characteristic warrants some additional examination. Cloud computing services are designed based on an assumption that the service provider will deliver capabilities to the third parties who act as consumers.

A core characteristic of cloud computing is the division of concerns between the cloud service provider and the cloud service consumer. The consumers of a service need only care about what a service does for them, not how it is implemented. However, to be successful, the provider must implement the service in such a way that it can deal with a global-class delivery environment and serve a wide range of consumers with a common service model.

In the early stages of cloud computing, consumers may need to examine implementation models to ensure that they are dealing with a strategically viable provider. Cloud computing services can be divided into five categories used as a foundation to build solutions plus a sixth category that deals with control and security.

System Infrastructure: Virtualized system software (such as server personalities, compute, storage and networking) on which the consumer can run any sort of application. Web-based provisioning — via a browser form, programmatic calls or automated response to an application load — is used to provide dynamic access to these resources. Minimal Web standards exist at the system infrastructure layer and provisioning models are generally vendor-specific.

Application Infrastructure: A set of services that parallels the traditional middleware and development technologies. Services must be built to exploit Web-centric architectures and global-class design principles. A new category called "application platform as a service" is emerging, with vendors providing global-class, Web-Centric Rapid Application Development (RAD) environments.

Applications: Applications that are designed for global class delivery and delivered as a service using web-centric architectures to a browse are considered cloud application services. This generally requires creation of a multi-tenant architecture in which a single application instance supports many organizations (each of which may have multiple users) but provides a unique view for each organization.

Customizations and extensions, as well as data, are isolated between tenants/organizations by default, but may be selectively shared (for example, salesfore.com). Cloud application services should be built using Web architectures and may expose components or interfaces for mashable access, control or extension.

Information: Content access and search services or Information services delivered as feeds via RSS/ATOM or other web models. 

Business Process: Any business process (including payroll, printing and e-commerce) delivered as a service via the Internet with provided via web-centric interfaces and WOA access mechanisms.

Management and Security: Services to manage the access, consumption, delivery, and service levels associated with cloud services. While these have some similarity to the operations management tools used inside an enterprise the demands of global class cloud based applications will require additional capabilities.

Cloud computing is heavily influenced by the Internet and web and the vendors that have sprung from this phenomena. Users evaluating cloud computing should focus first on the consumer/provider relationship and the nature of the services being provided. Evaluation of enabling technologies and their impact on the cloud services and/or their applicability to in-house infrastructure and application models is a related, but separate exercise.

A more sophisticated approach to running applications in the cloud would be for the consumer to build a solution on a global class application that is elastic, scalable and delivered as a service using internet and Web technologies. It may also uniquely exploit cloud-centric distributed and parallel processing capabilities. The resulting application would be considered a cloud application service.

Full exploitation of the cloud would use cloud based services to support the entire process of building, storing, running and managing the application. This approach is being championed by a new crop of RAD and database vendors that provide turnkey environments — aka application platform as a service (APaaS) — that build on the scalability of cloud-based infrastructures.

What services will be consumed via cloud/web platforms?

The most prevalent cloud-based application service is SaaS, where an application is delivered from the Internet/Web via a browser to an end user. Where these applications are built using Web architectures and expose components or interfaces for mashable access, control or extension, they are also part of the Web/cloud platform.

Web-based techniques and technologies (for example, REST, XML, RSS/Atom and Python), and vendor-driven (for example, Google GData) or community-driven (for example, Accord) standards, define a rich interoperability model. Extended data models and APIs driven by individual vendors emerge "at the edge" and collapse into common, open multivendor standards over time.

Multi-user applications delivered as a cloud service use a single application instance to support many users but provide a unique environment and experience for each user, including, where appropriate or desired, the ability to define the level of separation of data between users (for example, Google Apps).

Multitenant applications delivered as a cloud service use a single application instance to support many organizations (each of which has multiple users) but provide a unique environment for each organization. Customizations and extensions, as well as data, are isolated between tenants/organizations by default, but may be selectively shared (for example, salesfore.com).

Who will deliver cloud/Web platform services, and how should companies build/use them to support their business strategies?

One of the primary reasons why we believe there will be significant uptake of the cloud-based model is that cloud service providers can deliver significant economies of scale. Most organizations don't scale past tens of thousands of users, while cloud providers will ultimately provide services for millions of users.

Cloud providers are also working toward lights-out data center operations with extremely low-cost operations and reliance on copious amounts of inexpensive direct-attached storage (DAS). Enterprises, by contrast, will continue to run high-touch operational models with expensive storage area networks (SANs), which will consequently be allocated parsimoniously.

We expect the cloud providers to use replication services to make multiple copies of the data, and, thereby, avoid more-expensive redundant array of independent disks storage and tape backups, both of which enterprises will continue to use. Finally, we expect cloud providers to emphasize browser-based e-mail services that contribute to lower operational costs, in contrast to enterprises, which will continue to rely on fat clients running proprietary client-to-server protocols.

We expect cloud providers, such as Microsoft and Google, to create an offline browser client that will, for example, cache the past 30 days of e-mail.

The vagueness of cloud computing can be narrowed down to two basic "approaches." The first is an application of technologies, including virtualization and automation that focuses more on the "computing" than on the "cloud" aspect, with emphasis being placed on the technologies that enable the creation and delivery of service-based capabilities. The second is the "cloud" as an Internet/Web/Saas-originated idea, with credit often given to Google's Eric Schmidt for the term. The focus is more on "cloud" than "computing," with the emphasis being placed on access to services from outside of the enterprise (that is, ''from the cloud").

The author is SDG Head at Datacraft