Where information technology was concerned, the term "cloud" used to exist only on paper - in circuit diagrams and systems schematics, metaphorically representing an infrastructure for handling reliable interconnectivity (i.e., the network). Now, thanks to the many efforts to physically realize the mantra "the network is the computer," cloud computing is fast becoming a reliable infrastructure for handling a much broader spectrum of services, including massively scalable processing power and storage capacity.
Current trends forecast a rather sunny outlook for cloud computing. It is promising to be what business and IT executives (and their teams) have been asking for - a truly elastic and highly affordable computing infrastructure. When it matures, it will be able to totally free them from building IT infrastructure and keeping it running, and it will help them redirect most of their time and energy toward innovating and solving real problems of their core businesses. They will begin to take computing power for granted, just as they do their electric power.
Under those circumstances, on-premises computing would gradually become passé. Some market analysts estimate that three to four years from now, 20 percent to 25 percent of the IT workload will be managed using cloud computing environments.
Background
In its current form, this fast-moving disruptive paradigm began in late 2006 when a Google engineer, Christophe Biscigila, signed up to teach a class at University of Washington on "Google 101." He used the opportunity to teach students how to solve problems involving massive computer clusters and terabytes of data.
With the backing of his CEO, Bisciglia built a mini data center of 40 servers and leveraged Google's core algorithm, a quintessential piece of functional programming called "Map/Reduce," to demonstrate how computing could be exposed as a well-defined and well-packaged service.
Later, Google and IBM joined forces to create a "research cloud" in which six universities - the University of Washington, Stanford University, Carnegie Mellon University, Massachusetts Institute of Technology, the University of Maryland, and the University of California, Berkeley - have participated. Located in data centers at Google, IBM's Almaden Research Center, and the University of Washington, this cloud will eventually include more than 1,600 processors.
As of May 2008, both IBM and Google announced plans to go beyond the realm of academia to start offering commercial services involving cloud computing. Meanwhile, innovative businesses like Amazon.com have seen the opportunity to leverage their own massive computing facilities to create cloud computing offerings around them.
Amazon's AWS (Amazon Web Services) is the end product of this endeavor, and it includes the Elastic Computing Cloud (EC2) and Simple Storage Service (S3), both of which are off to a great start. Marquee clients like The New York Times and NASDAQ, as well as startups like Zillow.com and Animoto, are building brand-new businesses on the Web thanks to the low barriers of entry afforded by the cloud.
Context
Before we begin to analyze the technical qualifications of a cloud computing environment, it is important for us to first understand the evolving nature of this new paradigm. Cloud computing is closely related to grid computing and utility computing.
In grid computing, each computer is allowed to access and use the resources of every other computer on the network of a much larger pool of disparate computers. For cloud computing, the number of such computers is a little more controlled, and the computers only represent a homogeneous back end of the network.
Utility computing is more of a business model, where one company pays another company for access to computer applications or data storage. Elasticity, and the related on-demand provisioning and de-provisioning, are not necessarily features expected of utility computing.
Commercialization
IBM defines cloud computing as an emerging computing paradigm where data and services reside in massively scalable data centers and can be ubiquitously accessed from any connected devices over the Internet. The devices include the 4+ billion (and counting) wireless phones, personal PCs, TVs, and desktops and servers used in businesses.
The immediate commercialization of cloud computing would more or less categorize it as a "service" delivery pattern, only because that is the best way to extend its reach as something tangible and packaged. Consequently, we should not be surprised if the term "service" is a bit overused where cloud offerings are concerned.
Almost every aspect of computing will be packaged as a service, including hardware (HaaS Hardware as a Service), software (SaaS Software as a Service), or some combinations/derivatives thereof, such as platform (PaaS Platform as a Service), and capability (CaaS Capability as a Service). Each will have its own commercial flavor, corresponding features, and service-level agreements (SLAs).
Architecture
The candidate applications for cloud computing are practically limitless. With the right stack of software architecture, a cloud computing system can run all the applications of a normal computer.
In addition to the common layers of application software, the cloud comes with additional layers to support elasticity, provisioning/de-provisioning, monitoring, metering/billing, clustering/load-balancing, and related fault tolerance and recoverability. (See Fig. 1.)
Benefits
The cloud would universally level the playing field, bringing down barriers that stand in the way of smaller companies. All kinds of transactional and demographical data that is publicly available on the Internet can now be processed with the massive and significantly inexpensive power of the cloud and made into insights and services of dramatic value.
For any enterprise, the cloud can enable innovation and business transformation by acting as a highly collaborative platform for building incubation labs and innovation factories.
On the more tactical side, the cloud offers a low-cost and expedient model for acquiring IT services for consumers and enterprises alike. Some of the early adapters have taken advantage of these benefits to launch applications in a matter of weeks (as opposed to months) and for fractional costs, as compared to those incurred using in-house computing facilities, services, and processes.
On-premises computing has its limitations. For many next-generation applications, such as Extreme Transaction Processing (XTP), virtual worlds, policy-driven Service-Oriented Architecture (SOA), semantic Web, streaming video, real-time speech translation, and mobile commerce, the power of the cloud actually becomes a necessity. (See Table 1.)
Future
Bisciglia's idea of the cloud is definitely mushrooming and finding wider acceptance among users and providers. Yes, there are nagging concerns around security and privacy, with some that are truly technical and some that are basically part of the inertia of accepting a new paradigm. These concerns are being addressed through further improvements in virtualization and firewalls.
Provisioning is becoming more user-friendly through the deployment of Web 2.0 technologies and standards such as Elastic Computing Markup Language (ECML) and Elastic Deployment Markup Language (EDML). These standards are also being extended to include SLAs such as provider's reliability, scalability, and performance commitments.
The non-technical concerns will linger in business areas that are traditionally resistant to change and where paranoia is a virtue. But most of the business world will gradually come to grips with the reality of dealing with "virtual enterprises" that have distributed accountability of the computing power behind their products and services.
Advancements in enabling technologies will fuel the growth and acceptance of the cloud. Broadband access for wired and unwired connectivity, the move from multi-core to many-core chip technology - which is pushing up the limits on the elasticity of the underlying computing power - and the investments in parallel programming that will result in next-generation, high-speed infrastructure and application software are going to make the cloud the most compelling aspect of the Internet.
The network will truly be the computer, supporting us in our efforts to be innovators and problem solvers, and helping us to bring next-generation products and services to market faster and cheaper.
Sreedhar Kajeepeta is senior VP and CTO of Covansys, a CSC company. Contact him at skajeepeta@csc.com.
