Customization and Standardization: A View of Cloud and Grid Computing
Sequencing a genome, storing vast video libraries, or utilizing a non-essential application for occasional use are all functions performed within the realm of information technology. Meeting a particular task need was once the challenge of the end user or organization to ascertain their current and future use to guide their technology purchases. In the not so distant past, limitations of hardware or budget constrained the application of technology. Networking information technology proved one solution to sharing resources and boosting capabilities, while at the same time allowing for centralized governance models to facilitate access and protect privileged information.
As the capabilities of technology advance in computational power, storage, and connectivity, new uses emerge to enhance the capabilities of science, business, and individuals. Scalability, the matching of correct resources to a particular need in time, both up and down, has long proved an elusive target for users and organizations. As specialized organizations with robust computing networks and a common interest sought to enhance the capabilities, grid computing emerged to enable pooled resources.
The newest label for large-scale, shared resources is 'cloud computing.' References to 'the cloud' raise questions of whether this is a unique approach to delivering scalable resources, or merely a breezy marketing label for grid computing. Ian Foster et al. survey this question in a 2008 conference paper entitled, "Cloud Computing and Grid Computing 360-Degree Compared," (Foster, Zhao, Raicu & Shiyong 2008). Lacking a definite line of demarcation between grid and cloud computing is best characterized by noting the overarching goal of both, "the vision is the same -- to reduce the cost of computing, increase reliability, and increase flexibility by transforming computers from something that we buy and operate ourselves to something that is operated by a third party," (Foster, Zhao, Raicu, & Shiyong, 2008).
This paper provides an exegesis of the survey, by Foster et al., of Grid and Cloud platforms, as well as a focus on the particulars aspects that most concretely differentiate. The method of this paper briefly follows the course of Foster et al. through six areas. Through the business model, architecture, resource management, data models, applications, and security a clearer view of the similarities and differences of the Grid and the Cloud emerge. This paper refrains from long-term predictions of convergence upon a single grid or cloud-computing paradigm, but does note trends that fit with a more general theme of computing as a utility.
The most visible differentiating characteristic between cloud and grid computing lies with the business model. Foster et al. largely accept that the evolutionary course of information technology is comprised of three main functions: computation, data storage, and connective infrastructure. In his famous article, "IT doesn't matter," Nicolas Carr noted that, "the core functions of IT -- data storage, data processing, and data transport -- have become available and affordable to all," while arguing that a technology-based business strategy relied upon maintaining proprietary capabilities (Carr, 2003). Foster et al. note an egalitarian business model that integrates Carr's view in, "a cloud-based business model, a customer will pay the provider on a consumption basis, very much like the utility companies charge for basic utilities such as electricity, gas, and water, and the model relies on economies of scale in order to drive prices down for users and profits up for providers," (Foster, Zhao, Raicu, & Shiyong, 2008).
In contrast, Foster et al. note that the Grid computing business model, "is project-oriented," and rationed through a currency of "service units," (Foster, Zhao, Raicu, & Shiyong, 2008). Grid computing resembles a co-operative, in that joining integrates your node into a larger external network that enables shared use, and creates the experience of a virtual organization, that Foster et al. define as, "a logical entity within which distributed resources can be discovered and shared as if they were from the same organization," (Foster, Zhao, Raicu, & Shiyong, 2008).
Security & Architecture
While both represent large scale sharing of resources aimed at enabling a dynamic scaling of resources to match needs, a divergence of architecture and accessibility provides another key point of differentiation between the Grid and the Cloud.
If operational continuity is viewed as a security issue, then clearly the Grid retains an edge over the Cloud in the event of service or connection disruptions. Due to the fact that the Grid is comprised of smaller nodes that connect for resource sharing, the node retains core capabilities should an outage to the larger network occur. The cloud also retains the end user, or node, capabilities, however depending upon the type of services employed from the cloud may result in partial or complete disruption if an outage occurs.
A key characteristic of cloud computing is the configuration of services aimed at a very broad user group. The homogenous nature of a cloud operation is noted by capturing economies of scale and that, "the construction and operation of extremely large-scale, commodity-computer data centers at low-cost locations was the key necessary enabler of cloud computing," (Armbrust, et al., 2010). Mass appeal and scale entails a lower threshold of security and control, since little more than a credit card is necessary to gain access to most cloud computing platforms. Foster et al. note the easy access of the Cloud with, "Note that new users could use Clouds relatively easily and almost instantly, with a credit card and/or email address," (Foster, Zhao, Raicu, & Shiyong, 2008).
Grid computing is notably targeted at more specialized groups. Organizations heavily reliant upon information technology are most likely to have robust computing capabilities, budgets for specialized applications, and security of data is of the utmost importance. Grid computing may best be characterized as the model adopted by industry leaders and experts that benefit from shared resources, but only amongst their peers. The scientific community is more likely to be found decoding and sequencing genes in a grid configured system.
The cloud-computing model may be best suited for small business, whose computing needs are best served by specialized third parties whose core competency is information technology. A real estate office or mortgage-brokering firm is apt to find cloud services attractive as it outsources the expensive and onerous tasks that are not integral to the business. The counterpoint of this centralized and standardized approach is that security updates and monitoring is more conducive to this type of environment. Organizations that rely upon proprietary information and continuity of service are unlikely to fully embrace cloud services for the foreseeable future, however this merely denotes one area where customization is warranted for particular cases.
Application & Data Models
Foster et al. note various points of divergence amongst the Grid and Cloud amongst applications and data models, however none emerge as key points of differentiation. These areas are noteworthy for their current dissimilarities, but they are not fundamental.
While grid computing is comprised of various organizations and systems patched together, access is gained via, "standard, open, general-purpose protocols and interfaces," (Foster, What is the Grid? A Three Point Checklist, 2002). The standardization of protocols delivers the added benefit of interoperability across networks, and makes use of, "Grids focused on integrating existing resources with their hardware, operating systems, local resource management, and security infrastructure," (Foster, Zhao, Raicu, & Shiyong, 2008). The adoption of general-purpose protocols also limits widespread adaptation of newly developed applications. Furthermore, the disparate nature of Grid network nodes imposes bottlenecks on data transferring amongst heterogeneous machines and connectivity.
By contrast, cloud computing is noted for its homogenous product environment that, at present, offers no such similar portability of data as the Grid. Any customization of applications or interface must occur within the confines dictated by the Cloud platform's centralized control, thus, "Clouds mostly comprise dedicated data centers belonging to the same organization, and within each data center, hardware and software configurations, and supporting platforms are in general more homogeneous as compared with those in Grid environments," (Foster, Zhao, Raicu, & Shiyong, 2008). The constrained experience may be of no issue to most organizations, but is likely a key limiting factor for others. A key characteristic of the Cloud is the proliferation of data farms that deliver relatively faster access to data, however this remains an open question, as operations scale up, whether a similar bottleneck will be realized.
The issue of cloud inoperability across platforms and application development may be resolved with further development of Cloud "federations," however it remains that organizations reliant upon adaptability may best be served through the individual node autonomy of a Grid (Yang, Nasser, Surridge, & Middleton, 2012).
The Grid's most fundamental characteristics are the information and management protocols to allocate and queue resources. The Grid also benefits from a transparency of its accounting system that eludes the Cloud.
However, resource management of peak loads and heavy use within the Cloud's standardization enables a more fluid introduction of additional hardware with the simple installation of additional servers or data storage. The Grid's resources are generally expanded…