Chapter 3: Cloud Computing and the New Economics of Business Michael Hugos OVERVIEW Internet-based technology is driving economic change at a level not seen since the spread of industrial technology in the late nineteenth and early twentieth centuries. What became known as "Web 2.0" and the business and consumer applications it brought about have continued to evolve, and what has emerged is now known as cloud computing, software-as-a-service, and social media. There is as yet only a short history of using these technologies and they continue to evolve. So we have much to learn, but it is quite clear that they are leading to disruptive changes in the way we communicate with each other and in the IT infrastructures that companies use to support their business operations. The spread of cloud computing is an excellent example of the phenomenon known as "creative destruction," which was popularized by the economist Joseph Schumpeter.[1] Schumpeter pointed out that in capitalist economies, there are waves of change where the introduction of a new technology or new process for doing things upsets and replaces the previously dominant technology and the people and companies who used that technology. Cloud computing is having this effect on vendors who sell traditional versions of computing technology, and on the people who make their living operating this technology. Companies that have large investments in traditional in-house computing technology will not abandon those investments immediately, nor should they. The transition of companies to cloud-based technology will be quicker for some and slower for others, depending on their individual circumstances. But the change will happen. History shows over and over again that resistance to the spread of new technologies is almost always futile (and often fatal). People and companies that resist are finally forced out of business and replaced by others that do adopt new technology. Clearly the best strategy for people and companies is to actively explore the opportunities for cloud computing, begin appropriate projects to gain experience in its use, and understand its strengths and weaknesses. A COMBINATION OF TECHNOLOGIES CREATE CLOUD COMPUTING Since the turn of this century, several different but related kinds of information technology have been evolving rapidly, and they are now being combined to make it possible to deliver computing resources on demand to companies almost anywhere in the world. The combination of technologies, such as the Internet, Web browsers, server virtualization, parallel computing, and open source software, produces a whole new set of possibilities for delivering computing resources. The term "cloud computing" is now used to describe the result of combining these technologies. IT vendors are offering combinations of these technologies to companies that want to outsource some or all of their traditional IT operations such as running data centers and operating traditional application packages, like enterprise resource planning (ERP), customer relationship management (CRM), and other business support applications. SOME WORKING DEFINITIONS OF CLOUD COMPUTING The exact definition of cloud computing is still evolving. Cloud computing is both a model for delivery of business computing services and a method for managing and operating computing hardware and 1 | Chapter 3: Cloud Computing and the New Economics of Business software infrastructure. Different IT vendors put their own spin on their definitions, but they share more commonality in their definitions than differences. Here are several working definitions: • • • • "Consumer and business products, services, and solutions delivered and consumed in real time over the Internet" (IDC)[2] "A style of computing where scalable and elastic IT capabilities are provided as a service to multiple customers using Internet technologies" (Gartner)[3] "… a broad array of Web-based services aimed at allowing users to obtain a wide range of functional capabilities on a ‘pay-as-you-go’ basis that previously required tremendous hardware/software investments and professional skills to acquire." (Jeff Kaplan)[4] "… a way of utilizing resources wherever they may be when you need to use them. In that sense you just need to insure that your networking, security, and hardware infrastructure are robust enough to deliver the resources when needed, but just as important, your applications need to be able to execute well in that environment. To me, it is having what you want, when you want, through your virtual desktop no matter where you are." (Frank Enfanto)[5] From these definitions and lots of other definitions (do a Web search on "cloud computing definition"), there are three characteristics that everyone seems to accept when it comes to describing cloud computing. Everyone agrees that cloud computing has the characteristics of: 1. Practically unlimited computing resources. Resources such as computing power, data storage space, and additional user sign-on IDs for applications are available on demand as needed, and this enables a high degree of agility and scalability in meeting evolving business needs. 2. No long-term commitments. Computing resources are immediately available and they may be used as long as needed and then retired because they are acquired on a month-to-month or even a minute-to-minute basis. 3. Pay-as-you-go cost structure. Because there are no long-term commitments, the cost of cloud computing resources is a variable cost, not a fixed cost; cost fluctuates depending on the amount of usage. CLOUD COMPUTING HAS THREE COMPONENT LAYERS The pace of change in cloud computing technologies is rapid. Certain components are changing so fast that the names and technical details of how they operate alter significantly every six to twelve months. Nonetheless, it is possible to group cloud computing technologies into three basic categories or layers. These layers support each other, and the relationships between the layers and the way each layer operates are relatively stable. We will use these three layers to create a basic model of cloud computing and provide a stable framework to discuss cloud computing technology (see Exhibit 3.1). The three layers of technology used in cloud computing are: 1. Hardware virtualization 2. Data storage and database management 3. Applications and application development environments Hardware virtualization is a term that refers to the abstraction of computer resources so that many different computers or application servers seem to be available to run different application system, even though there may be a much smaller number of physical computers actually present. The term "virtual 2 | Chapter 3: Cloud Computing and the New Economics of Business machine" (VM) refers to a software implementation of a computer or application server that executes programs like a real physical machine. Hardware virtualization enables companies to optimize the use of physical server resources and improve server administration. Virtualization is a common practice on mainframes and is becoming widely available for other computer architectures, such as computer servers built from low-cost computer chips and commodity hardware. In the cloud computing world, this layer is also referred to as infrastructure-as-a-service (IaaS). Exhibit 3.1: Three Technology Layers of Cloud Computing Data storage and system management based on hardware virtualization is far more efficient and flexible than was possible before. Instead of buying a new physical server to host each different database, different databases can be supported by different virtual machines. The processing power of these virtual machines and the storage capacity of these databases can then be dynamically changed based on actual business requirements as they occur. In addition, these different virtual servers can be set up to run different operating systems, such as Linux or Windows, as needed. In the cloud computing world, this layer is also referred to as platform-as-a-service (PaaS). Applications and application development environments can leverage hardware virtualization and data storage and database management capabilities in a cloud computing environment. Application systems to support different business operations can be hosted on virtual machines that are scaled up or scaled down hour-by-hour as needed to meet business user demands. New copies of a given application system can be created instantly and put into operation as needed. In cloud computing this layer is referred to as software-as-a-service (SaaS). Application systems can be developed on cloud computing platforms that support different programming languages, testing platforms, and system management tools, depending on what system developers wish to use. Google, Amazon, and Microsoft provide some popular development environments. Popular programming languages that are supported are languages such as Java, PHP, Ruby on Rails, and C#. IMPLICATIONS OF THE TRANSITION TO CLOUD COMPUTING The momentum created by the pressures of our present economy is driving us inexorably toward cloud computing. Larger companies are creating their own internal clouds and smaller companies are moving to clouds from external service providers. The twin concerns of performance and security are valid, but they are not excuses or reasons to prevent the switch to cloud computing. Vendors are rapidly delivering tools to respond to and manage these concerns. 3 | Chapter 3: Cloud Computing and the New Economics of Business The move to cloud computing is the most profound change in IT since the emergence of the Internet. Big changes are occurring in the mission of in-house enterprise IT organizations and the way they are run. The bulk of staff in these organizations are presently devoted to performing traditional IT functions such as operating and maintaining data centers, data networks, and PCs, as well as the monitoring and enhancing of application systems that are hosted in those data centers or running on desktop PCs. Adoption of cloud computing will shift most of these traditional activities out of enterprise IT groups and into the cloud service provider organizations. The information technology profession as we have known it for the last 40 years is dying; its obituary is already written. Companies are transferring the risk of high-ticket technology investments—like wholly owned data centers and internal application hosting—to highly focused and specialized service providers. Who's looking after the network? A service provider. Who's monitoring application performance? A service provider. With the move to cloud computing, in-house IT professionals in most enterprises face big changes in their careers and risk to their job security and earning power. The spread of cloud computing is creating disruption on a level commensurate with the degree of change it ushers in. Just as some IT professionals in the 1980s resisted introduction of personal computers in their companies, and some IT professionals questioned the value of the Internet in the early 1990s, some IT professionals now resist the introduction of cloud computing in their companies. New technologies like these reduce demand for certain traditional skills, and they change the way the IT profession is organized. Cloud Performance and Security Concerns As it was one hundred years ago with the debate about whether to rely on outside vendors to provide electric power, the two main concerns in today's debate about cloud computing revolve around issues of performance and security. Because of the demand for new IT operating models created by economic imperatives of our global real-time economy, there are many technology vendors creating products to address these performance and security concerns. New startups and established IT vendor companies are developing products that provide performance monitoring tools for cloud computing environments. Cloud computing service providers are purchasing these performance monitoring products in order to support the growth of their cloud computing businesses and to assure their customers that they can monitor performance and deliver consistent and high levels of service. For many application systems there are already adequate performance management tools available. In other cases there are still significant technical issues to be addressed, but if the history of technical development in the last several decades is a guide, we can safely assume there will be rapid progress made on these issues in the next few years. IT vendor companies are rolling out suites of new products that enable companies to provide better security for cloud computing environments. They offer products for cloud intrusion prevention and for global threat correlation. Using these products, companies can create computing and collaboration environments that integrate in-house IT infrastructure with cloud-based application systems, and they can exercise a high degree of control over who enters those environments and what information those people have access to. 4 | Chapter 3: Cloud Computing and the New Economics of Business As these products rapidly improve, they become analogous to good brakes on a racecar; the better the brakes, the faster you can drive the car. Good performance monitoring and security protection enable companies to go faster and faster in setting up new cloud computing applications because they don't have to worry about performance and security concerns that would otherwise slow them down. Cloud Computing Drives Creation of New Businesses Under the relentless pressures of economic necessity and unpredictable market conditions, companies simply have to find ways to shift the cost and risk associated with basic IT equipment and operations to outside vendors. These vendors can then amass huge demand for their services, make the investments, and build the data centers needed to create new economies of scale that deliver computing services at lower and lower price points. Cloud computing data centers are evolving into factories that supply computing power, data storage, and application systems to drive the rest of the global economy. Plans to simply cut IT budgets and try to keep operating expenses down until business rebounds will not work. If companies restrict IT operations and IT is seen and used only as a cost center instead of part of what creates and delivers the company's value proposition, then a company will not be able to roll out new products in a timely manner or keep up with changing desires of its customers or respond quickly enough to new threats and opportunities. It will not be a winning move for business or IT executives to take a position analogous to the in-house electric power people from one hundred years ago. Instead, a far better move is to find ways to be part of enabling the transition to the cloud and moving your enterprise toward a more variable cost operating model. There are many opportunities for people who can show companies how to move to cloud computing and, at the same time, how to address the performance and security issues that go along with this move. When companies make this move they will have much more money to spend on doing the things their customers pay them for and creating the evolving stream of products that keep them connected and relevant to their customers and earn them profits. This shifting of functionality to outside service providers must happen in order for in-house IT organizations to redirect their time and money toward working with the business units to use IT as part of what creates and delivers a company's value proposition to its customers. IT in the form of social media is fast becoming the way that companies spread the word about their products; traditional print advertising is losing its effectiveness. IT in the form of customer support and relationship management systems is how companies connect with their customers and build long-term relationships. And many companies produce products such as financial services, consumer electronics, smartphone and Internet applications, and entertainment and consumer services of all sorts that literally are built and delivered as IT themselves. A useful analogy to think about is this: Universal access to low-priced electric power made possible by the spread of electric utilities drove a wave of innovation, not only in how businesses operated but also in the products they developed. From the 1910s onward, the introduction of thousands of new products using technologies such as electric motors and vacuum tubes became possible because everybody had dependable electric power. How many companies were created to build and sell products built with components like electric motors and vacuum tubes and transistors? What business innovations and new products can you now imagine based on universal access to low-priced cloud computing power? How many new companies will be created to develop and deliver those products? 5 | Chapter 3: Cloud Computing and the New Economics of Business A BUSINESS STRATEGY BASED ON AGILITY This section and the rest of this chapter expand on agile IT system development methods and agile IT architecture principles, discussed in Chapter 3 of the first edition of CIO Best Practices. Cloud computing can be considered primarily as a cost-saving technology, and can be used here and there on cost cutting projects and quick fixes to provide point solutions for certain operational problems. Or cloud computing can be understood in the context of an overall business strategy based on agility and responsiveness. Cloud computing certainly provides cost savings in some situations but cost savings is not the most important benefit. The real value of cloud computing is the way in which it can be used to support a strategy of business agility. A strategy of business agility puts responsiveness before efficiency. It emphasizes the ability to make continuous incremental changes and adjustments in operating procedures to respond as the world unfolds, so as to best fit new conditions. It also emphasizes continuous exploration of new business opportunities followed up by rapid growth into new markets when exploration shows a business opportunity to be profitable. An Example of Business Agility Here is a case in point; suppose a company, GrowMore Corporation, spots an opportunity to leverage its existing expertise and supplier relationships to launch a new product line for a market adjacent to its traditional markets. In order to do this, the company wants to set up a new business unit with branch offices in key geographical locations. It wants to have sales offices in these areas, and it wants to support the sales staff in these offices with a CRM system that enables sales-people to prospect for customers, create presentations and proposals, and follow up with prospective customers in a timely and organized manner. GrowMore Corp. also wants to collect sales and prospecting information from all the regional offices and store it in a single database at headquarters to provide for overall reporting and tracking of sales and business development activities. The new product line will need some customization for individual customers so that it best fits the customer's unique needs, but GrowMore Corp. does not want to staff all these offices with engineers to do this customization. Instead of sending engineers out with salespeople to make calls on prospect companies, the company wants to set up videoconferences. This way their salesperson on site with the customer or prospect can arrange for videoconference interaction between an engineer and a customer, and the engineer can interview the customer and collect the information needed to configure the product. In the old days, management of the new startup business unit would have submitted a support request to the company IT organization. The IT organization would then have sent out a business analyst to evaluate the request and study the needs of the new business unit. Then the request would have been prioritized against requests from other business units, and since existing business units typically get priority in the allocation of available IT resources, the startup unit would likely have to wait until the next budget cycle before it could get funding for the IT services it needs. Then, when the funding and IT resources became available, there would be a process of designing and developing the needed software or evaluating possible packaged software solutions. Then there would be the purchasing and installation of the needed hardware and the communications networks, and 6 | Chapter 3: Cloud Computing and the New Economics of Business finally, the rollout of the new system and accompanying user training. During that time, months or even years would have passed. In many cases, the window of opportunity for the new business would have closed, and the solution delivered would be too late or would not effectively address business needs that had evolved and changed during the time it took to build and roll out the system. Instead of all this, management of the new startup business unit can go directly to relevant cloud services providers and start using one of their SaaS offerings within a matter of a few hours or a few days at most. Unlike the old days, there will not be any big upfront cost involved (no capital expense allocation needed) and the cost of operating the systems will vary, depending on the amount of usage. That way, there is no sunk capital cost if the business idea doesn't work out, and there is not much in the way of system operating costs. If the new business does work out, then operating costs can be paid for out of sales revenue. Using cloud technology and SaaS applications, the opportunity exists for this new business unit to put together a system by combining SaaS applications for their CRM needs and their video conferencing needs. They can add other industry-specific SaaS applications to their systems infrastructure as needs evolve. Or the new business unit can develop some unique custom applications as justified by the value proposition, and those applications can still run in the cloud, so there is no need to buy computer hardware and staff up the new business with IT people to operate that hardware. Implications of Cloud-Enabled Business Agility Cloud-Enabled Business Agility lowers the cost of sales for a new business unit in a way that then makes it possible and profitable for the unit to pursue smaller deals that were previously not profitable. The company can build a base of business from many smaller deals, which may be easier to get instead of having to go after only the larger projects and fight all the competition that goes along with getting those larger deals. These are some of the benefits of using an agile business strategy and the technology that enables such a strategy. The case study later on in this chapter further illustrates these ideas of agility. Cloud computing has great cost cutting potential in certain situations, and at the same time, it's important to keep the larger business strategy in mind. The agility benefits far outweigh the purely costsaving benefits. If an enterprise strategy calls for improving its ability to bring new products to market and its capability to expand geographically and open new offices, then cloud computing is a powerful technology for attaining these objectives. Using cloud technology to enable new business formation and new product development creates what could be called multi-national small and medium businesses (SMBs) because with cloud technology, SMBs can now be truly global, where 10 years ago they could not afford the IT infrastructure to support global operations in an integrated fashion (see Chapter 5 for specific examples of this trend). For instance, small and medium sized businesses can open new sales offices in countries around the world, and conduct local sales campaigns in a way unthinkable before cloud computing. Opening new country sales offices simply means paying for more people to use a cloud-based CRM package and a cloud-based teleconferencing system. Specialists at company headquarters can back up salespeople with in-depth technical support without having to fly people around the world to meet with clients. USING THE CLOUD FOR BUSINESS ADVANTAGE 7 | Chapter 3: Cloud Computing and the New Economics of Business Companies need IT infrastructures that enable them to operate more efficiently and accommodate continuous, incremental changes in business operations. To that end, many companies are already using server virtualization, and some are also using service-oriented architecture (SOA) to better leverage their existing IT investments and get additional flexibility and responsiveness from their existing systems infrastructure. Companies are now at the point where they need to move on from an internal focus directed at maximizing use of IT resources to an external focus on supporting collaboration and new product development through use of cloud computing. Companies are moving from internally focused SOA projects to externally focused Web-oriented architecture (WOA) projects, where they begin using SaaS applications and combining them with internal applications that support collaboration with other companies to drive their growth. This will happen because cloud and SaaS vendors are becoming more and more like utilities, offering reliable computing power and basic applications like email, ERP, CRM, and a growing array of industryspecific applications. Over the coming years, these vendors will develop economies of scale and expertise that enables them to offer their services at a much lower cost than what most companies would spend to deliver those services internally. Because of these developments now and over the coming years, companies will outsource more and more of their basic IT operations to manage their costs for basic IT services. This will, in turn, enable companies to shift more of their time and attention to doing things with IT resources that add value to their products and provide meaningful differentiation in the eyes of their customers. IT will be used to deliver competitive advantage. Cloud computing thrives in entrepreneurial environments where leapfrogging the competition is a daily motivator. Innovators need tools that fit their fast pace, their work-anywhere mentality, and their collaborative instincts. Cloud computing sets the stage for corporate innovation. Freed from lengthy implementation projects, moribund legacy applications, and armies of consultants, IT personnel can turn cloud computing into a competitive advantage. Cloud computing offers significant advantages in its low startup costs, quick delivery of computing resources, and its pay-as-you-go cost structure. In addition, it offers ease of management, scalability of systems as needs grow, and device and location independence so people can access these systems from many different devices, such as a PC or a smartphone like a Blackberry or iPhone. And finally, cloud computing enables rapid innovation in companies to respond to evolving markets. Many SaaS vendors and cloud service providers come to enterprise IT from the consumer IT side, and that means they are already focused on providing that customer-friendly interface that makes their software and services easy to learn and use. They bring the simplicity developed through their consumer experience, apply it to the corporate world, and continuously integrate with mobile devices such as Blackberries, iPhones, netbooks, electronic book readers, and iPads. These mobile devices are quickly becoming the new interface between people and the online world. BUSINESS APPLICATIONS WITH THE GREATEST POTENTIAL 8 | Chapter 3: Cloud Computing and the New Economics of Business Apply a pain-versus-gain measurement to determine which applications could work well in the cloud. If your company is a startup operation then, almost by definition, most applications will be good candidates for cloud computing because of the advantages described earlier in the GrowMore Corporation case study. If you are starting with an established company and existing infrastructure of in-house systems, then applications with the following characteristics are good candidates to start exploring cloud computing: • • • • • • Standalone applications with a low business risk if something happens and the system goes down or if system data were compromised or stolen Applications that are expected to have highly volatile and hard-to-predict workloads Situations where there is a need to collaborate and share information with an extended value chain of business partners Applications where there is a need to perform periodic data analysis on high volumes of data A platform to try things out quickly and at low cost, such as to field test a new application system or to create test and development environments for building new systems Situations where there is a need to conserve on capital expenses An example of a standalone, low-risk application is a system like a simple Wiki blog site to support information sharing and knowledge management within a company. Another example is a system that allows people in field offices to collect and share data and update business planning models or a system for use by the human resources group to pilot a new recruiting process. An example of applications with high volatility and usage patterns that are hard to predict is when a company launches a new product or a new product promotion and puts up a Web site to support that effort. If a site will only be up for six months or so and be prone to high spikes in traffic volume, then why spend the money and resources to sustain the site indefinitely, and why tie up systems infrastructure to support the peaks in usage volumes when that infrastructure will be underutilized during the non-peak times? For instance, a fast-food restaurant chain might decide to promote special low-priced value meals by establishing a Web site where customers can check for specials and see the location of nearby restaurants offering these specials. The site would be heavily accessed when certain specials are offered, but it is hard to predict the actual usage volume. In such a case the company could employ a cloud services provider to set up the Web site. The service provider could quickly set up the Web site and provision that site with computing power and data storage capacity as needed. This way the restaurant company would only pay for the actual usage it incurred, and its usage costs would drop during periods of low customer activity. Clouds also make sense where business is conducted based on shared data and where rapid feedback is needed. An example of this need to collaborate and share data with an extended value chain of partners could be a healthcare company that wants to share data on patient care and outcomes with a network of pharmaceutical companies and medical service providers. The healthcare company can make the data anonymous by blanking out the names of patients and then load the data into a cloud-based system where all relevant parties can access the data and apply cloud-based analytics to sift through the data for important patterns and trends. 9 | Chapter 3: Cloud Computing and the New Economics of Business Applications where there is a need to perform periodic data analysis on high volumes of data are also good candidates for cloud computing. For instance, an energy company doing geographical analysis to search for new oil fields could set up the compute-intensive portions of this operation using a cloudbased system. When large databases of information come in from its field exploration units, the company could ship this data up to the cloud where it would be processed and the results returned for in-house analysis. The company could strip out location-related data, so even if the data were stolen, it would not reveal sensitive information about where a promising new oil field might be located. In this example, the company would only pay for the computing resources used for the job when the job was run without tying up money in idle infrastructure when there were no jobs to run. Many companies are already using cloud solutions (IaaS and PaaS) to quickly provide their in-house application development groups with testing and development environments. Instead of going through all the expense and time of purchasing the hardware and software needed to develop new systems, the development groups of these companies can get what they need immediately and pay for it only as long as their need lasts. As one senior enterprise architect at a major corporation put it, "You can buy a gift card on Amazon and use it to set up your own data center." Cloud service providers such as Amazon, Google, Hewlett-Packard, IBM, Microsoft, Rackspace, and others offer immediate provisioning like this. CLOUD RISK CONSIDERATIONS There are a number of issues to consider when deciding which applications to put in the cloud and which cloud service providers to use. Those issues and their related risks tend to group up into the areas of system and data security, performance management and service level agreements, and vendor lockin. The issue of system and data security is perhaps the most frequently discussed IT risk. There are risks that data placed in the cloud can be compromised or stolen by third parties. Yet it is also important to see these risks in their proper context: the state of security that presently exists within many companies. Many companies have data security issues that are equal to or even greater than those faced by cloud service providers. In-house data center operations are cost centers, they are always under pressure to cut their costs, and that means all their security issues cannot be addressed. Data center operations and security are part of the way that cloud service providers and SaaS vendors earn their profits, so these companies are far more inclined to invest in security. Data and system security are issues that these provider companies have been working on for some time, and because their systems live in the cloud outside corporate firewalls, they get attacked thousands of times every day. This experience gives cloud service providers more security experience and better security procedures than that enjoyed by the average corporation. CIOs and CTOs should also remember that the greatest security threat to systems and data, whether behind an enterprise firewall or in the cloud, is something known as "social engineering." Social engineering refers to the various practices used to interact with system users, data center operators, and help desk staff to illegally gain access to passwords and user IDs. There are many ways that people can be tricked into giving away passwords and IDs to unauthorized parties who then use this information to gain access to the systems and data they want. Enterprise IT organizations are just as vulnerable to social engineering as cloud service providers. Protection against social engineering is more 10 | Chapter 3: Cloud Computing and the New Economics of Business a matter of managing human nature and maintaining effective policies for administering system access information. Performance management of cloud-based systems is an area that can make in-house IT staff uneasy. They are used to having more direct control over the actual computer, communications hardware, and operating systems that drive the systems their companies have used up until adopting the cloud. They worry that once systems are moved to a cloud environment, there will be no way to monitor and control the user response times and other performance characteristics of those systems. They feel that their companies will just have to accept whatever performance levels the cloud services vendor may offer and make the best of it. Since high levels of performance and satisfied customers are central to the profitability of cloud service providers, companies are investing in technology that allows customers to monitor and manage many of the operating parameters of cloud-based systems as discussed in Chapter 1. Cloud service providers are working with technology companies (Akamai, Cisco Systems, F5, IBM, and Nimsoft to name just a few) to design and install technology that monitors and displays real-time information, showing how well a given cloud application is performing. This technology enables in-house IT staff to respond and make adjustments as system slowdowns and other problems appear. Performance management technology will continue to improve as more and more use is made of cloud systems. Technical problems have technical answers, and technical advances can happen quickly when there is sufficient demand. Performance management is a justifiable concern of cloud customers, but it is not an obstacle that should prevent companies from making effective use of cloud systems. SLAs that guarantee certain levels of system performance are constantly evolving. At present, cloud service providers and SaaS vendors do not offer strict guarantees on their service levels. When service outages do occur, the vendors offer to reimburse customers for the cost of their services during the period of the outage, but they do not pay penalties that would reimburse customers for loss of business revenue or costs they incur because of a service outage. Again, it is important to see the issue of SLAs in the proper context: comparing the performance records of cloud service providers and SaaS vendors with the performance records of in-house data centers. In many cases there is no formal measurement of the service levels provided by in-house IT organizations. In-house systems go down or their response time slows down, and people in a company just accept that as a fact of life. The quality of in-house systems performance is directly related to the sophistication and training of inhouse IT organizations. While large enterprises can afford large and well-trained IT staffs and sophisticated data centers, many (maybe even most?) enterprises get by with underfunded data centers, poor automation, and understaffed IT operations. IT staffs at these companies do the best they can with what they have, but they also make no guarantees about the quality of their service or the reliability of uptime for their systems. For these companies, a cloud provider may be an improvement in quality of service. Lock-in is another risk in cloud computing that companies need to assess. Cloud computing often uses a different systems architecture than that used by traditional in-house systems. So once a system is put in the cloud it is not a simple matter to bring it back in-house or to move it to another cloud. Before 11 | Chapter 3: Cloud Computing and the New Economics of Business selecting a cloud provider or a SaaS vendor, it is important to evaluate the stability and longevity of that company, understand their pricing model, and project the likely ongoing costs associated with using their applications to run your business. Once again, it is important to see the issue of cloud provider lock-in by using an appropriate context: Lock-in already exists with traditional in-house systems and software. Once a company makes a commitment to use an ERP system and installs the software, there is a large degree of lock-in that comes along with that decision. It is expensive and time consuming to install an ERP system, and once done it is very unlikely that a company will go to the expense of uninstalling that system and switching to a different ERP system. The same goes for making a commitment to build company computing and communications infrastructure. Once the technology from a given vendor or small group of vendors has been installed, it would require a great effort to switch to the technology of different vendors. Over time it may actually be easier to switch from one cloud provider to another than it is to switch from one software vendor to another for your in-house systems. This is because clouds, by definition, are able to create virtual computers that you use to run your applications. As long as the operating system used by the virtual computers of one cloud provider is the same as the operating system used by another cloud provider, then it is possible to move the program code from one provider to another and expect that application system to work. Many cloud providers use a version of the Linux operating system. Other cloud providers use the Microsoft Windows operating system. Moving data from one cloud provider to another is the other issue to consider when thinking about moving your application system to a different cloud provider. Cloud providers such as Google and others are endorsing open standards (such as HTML 5.0) that make it easier to move data from the databases of one provider to the databases of a different provider. These same data transfer standards are also making it easier to move large amounts of data back and forth between in-house systems and cloud systems. CLOUD COST CONSIDERATIONS The cost of using cloud systems is often compared to the cost of buying and operating the system hardware and software in-house, and the answers provided depend on the depth of the analysis that is done. On one level, a person can simply compare the rate for renting the use of a virtual server from a cloud provider and compare that with the cost of buying a real server. After a certain number of months it would appear cheaper for a company to own and operate that server in-house. Therefore, if the application system it will power is expected to be used longer than that number of months, it might be more prudent to build and operate the system in-house rather than in a cloud. A senior enterprise architect, Rick Pittard, who works at a global 100 corporation, has been investigating this issue and working with some cloud-based applications to learn more. Assessing his corporation's cloud service provider, he observes that, "Hardware costs for short term projects, up to two years, are less expensive than purchasing and operating our own. Systems that will operate for longer than two years may be more cost effective to operate in-house."[6] Then he went on to add another insight, "but at three years you are getting close to replacement of the equipment, and if you go through the replacement cycle to upgrade your hardware, then there is no cost advantage to running it in-house; you might as well use a cloud provider instead."[7] 12 | Chapter 3: Cloud Computing and the New Economics of Business People often forget to add all the indirect costs that go along with purchasing, installing, and operating their own computer equipment. The indirect costs add up to much more than the purchase price of the hardware. In their cost analysis, people need to include all of the related costs such as: • • • People and electricity. What is the cost of the people in procurement who negotiate purchase prices and support contracts; what are the costs of the people in IT who operate the equipment; and what are the costs of the electricity and air conditioning and rent and operation of the data center where this equipment is located? System administration and asset tracking. What is the cost of the people who maintain the system databases and do the system upgrades; what is the cost of the people who manage the software licenses and hardware leases; and who then dispose of these assets at the end of their useful lifespan? Opportunity cost. What are the costs of not doing other things with the money spent on the above two areas? Are there other places where you could use that money for better return? About 70 percent of most company IT budgets goes to maintenance of existing systems and infrastructure. Maybe by using cloud services, a company can reduce this percentage and spend their money on things that produce a higher return. None of these costs address what may be the biggest cost of all: the cost of senior management time. If senior management in IT and operations spend most of their time on operations staffing, data center build out, and equipment leasing and installation, is that a good use of highly paid management time? Maybe it would be better if they spent more time figuring out how to use technology to sell more of the company's products and services, reduce operating expenses, and use IT for competitive business advantage. CASE STUDY: SELLING "Designer Chocolates" This company makes well-known chocolate candies and sells them through a variety of retail channels. It sells a lot of candy, but profit margins on candy are always being squeezed. Then some smart marketing people in the company spotted a business opportunity to sell "designer chocolates," cookies, and drinks through cozy storefront locations in upscale neighborhoods. The company won't sell as much this way as through traditional channels, but it will get a much better profit margin. It is an opportunity for the company to supplement its traditional business with a new business that can generate big profits for some period of time, but no one knows for how long. What would you do if you were the CIO or the COO and you were asked how you were going to support this new designer chocolate business? Here's what you could do; here's how you could create an agile IT architecture and leverage it to move quickly and support this new business venture. Imagine that Exhibit 3.2 is what your existing infrastructure looks like; it was created over the years to support your traditional manufacturing business. The key to meeting the company's needs for launching the new business is to leverage existing systems as much as possible, to hold down costs and speed up delivery times for new systems. You could use server virtualization to better utilize existing servers and avoid having to buy any new hardware. You could also use an open-source, bare-bones store point-of-sale system to support basic store operations. 13 | Chapter 3: Cloud Computing and the New Economics of Business Exhibit 3.2: Case Study: Selling Designer Chocolates You could set up a simple network using IaaS at each store to connect cash registers and PCs to the Internet. Using that connection and employing agile IT system development methods, you could use SOA to hook in functionality from the existing inventory control system to manage store inventories and use the existing ERP system to handle accounting and financial reporting. A new supply chain database (a data warehouse) could be created using PaaS to store and report on all the business transactions related to store operations. This would provide the data needed to learn and continually adjust and improve operating processes of the new business. Exhibit 3.3 illustrates this approach. You deliver the first version of the Store Support System needed to open a few stores to test the concept, and guess what? The concept is taking off! Business is good, and now the folks in marketing and sales want to open up more stores and add some new features and products to the business model. Once again, you are asked to deliver the systems capabilities needed to make the expansion possible. But still, nobody knows how far this expansion will go or how long this business will last. Exhibit 3.3: Designing the Network How would you use agile IT architecture to keep supporting the growth of this business? Given the uncertainty of the business, it's probably not wise to buy more servers to support more stores, because you would have to take on the cost and the risk of building out your data center, adding more system backup capability, and hiring more staff. Instead, you could use a cloud services provider to deliver all the computing power for the stores on a pay-as-you-go basis. This leaves you free to cut back on computing services if the business were to take an unexpected turn and not grow as expected. You could also combine the needs of the new business with needs of the existing business and look at retiring older IT architecture in favor of using more cloud computing and SaaS to meet changing company needs. This would turn fixed operating costs into variable costs and reduce any need for capital to purchase IT infrastructure. Operating costs would rise somewhat as business grew, but operating costs would also drop if the business did not grow as expected, so cash flow is better 14 | Chapter 3: Cloud Computing and the New Economics of Business protected. The company would not have to incur the risk of a big investment in IT infrastructure at this stage when the business is going through significant changes and long-term needs are hard to see. Exhibit 3.4 illustrates this approach. Exhibit 3.4: Augmenting System Agility These diagrams illustrate how a lot of companies are going to evolve their systems architecture in the coming years. Using these techniques and technologies enables companies to move quickly, yet also minimize their investment risk in case a new business doesn't pan out. These approaches are stable and scalable. They enable an organization to move quickly. Ready or not, this is what the future of responsive enterprise IT infrastructure looks like. DESIRABLE CHARACTERISTICS OF NEW IT ARCHITECTURE In the present economy, companies are looking for ways to cut IT expenses, yet the real opportunity is to find ways to manage total company expenses in such a way that they track with the demands of business operations. Saving 10 or 20 percent on a company's IT budget is relatively small savings compared to using IT to save 10 percent on the operating expenses of the whole company or using IT to grow company revenue by 10 or 20 percent. Companies have the opportunity to power their business operations with systems infrastructure that meets three operating standards: 1. Low capital expense 2. Variable cost of operations 3. Scalable computing platform Low capital expenses are the order of the day in business because revenue and profits are under pressure, credit markets are tight, and loans are harder to get, so there is less money for capital investments. Also, because we're in a period of rapid technological change, making big investments in technology is risky because it might result in your company investing in technology that becomes obsolete a lot faster than expected. Best practice CIOs are learning to get systems in place without a lot of upfront capital expense. Smart executives are learning to shift their investments from the building of facilities like data centers to delivering new business operating capabilities instead. Committing to the standard of a variable cost operating model is very smart because it's a great way to protect company cash flow. Pay-as-you-go operating models mean operating expenses will rise if business volumes rise, but just as important, operating expenses will drop or stay small if business volumes contract or don't grow as big or as fast as expected (you only pay more if you're making more 15 | Chapter 3: Cloud Computing and the New Economics of Business and you pay less if you're making less). In this economy, where it is so hard to predict what will happen next and where companies need to keep trying new things to find out where new opportunities lie, variable cost business models are best for managing financial risk. Committing to scalable systems infrastructure enables companies to enjoy the benefits of the first two standards. A scalable systems infrastructure enables a company to "think big, start small, and deliver quickly." Company executives can create strategies with big potential and try them out quickly on a small scale to see if they justify further investment. Companies can start with targeted 80 percent solutions to address the most important requirements first, then build further features and add more capacity as business needs dictate. PUBLIC CLOUDS, PRIVATE CLOUDS, AND HYBRID CLOUDS Although the exact definition of cloud computing is still being debated and different parties are offering slightly different visions, the cloud computing model has a handful of characteristics that are always present: massive scalability; provisioning of computing resources on demand; a pay-as-you-go cost structure; multiple systems and multiple users supported on the same computing infrastructure; systems and data available from anywhere with Internet connectivity; built-in disaster recovery; and software oriented toward customer ease of use. Three models of cloud computing can be defined within this broad vision of cloud computing: 1. Public clouds are owned and operated by third parties and located in data centers that are outside of companies that use them. Multiple companies share the resources of these cloud computing data centers; they are each assigned their own virtual resources based on a common set of physical resources. Public clouds are provided by companies such as Amazon, HewlettPackard, IBM, Google, Rackspace, and Salesforce.com. 2. Private clouds are owned and operated by a company or a cloud computing provider, but they are built for the sole use of a single company. Private clouds utilize the same technology as public clouds, and they are often built to enable an individual company to maximize the use of its computing resources and be more responsive to company needs than was possible under the traditional IT operating model. 3. Hybrid clouds are combinations of multiple clouds that are both public and private. These clouds are created by individual customers to meet their specific needs. For instance, a company may decide to create a hybrid cloud to combine a CRM system provided on a public cloud operated by SalesForce.com with an ERP system running on their private cloud, and they may further extend this hybrid cloud by combining it with the Google cloud to provide their employees with the collaboration and productivity tools provided by Google Apps. Hybrid clouds like these sometimes use the services of a cloud aggregator. Up to this point, this discussion has focused on the use of public clouds and the business reasons for using them. For many startup companies it makes sense to start immediately with the use of a public cloud, instead of investing precious capital in building their own data centers. By doing this, they avoid the distractions of running commodity computer hardware and software, and are able to concentrate on developing their unique value-added product or service that will be the profit generator for the company. 16 | Chapter 3: Cloud Computing and the New Economics of Business But for existing companies that have already made significant investments in systems infrastructure that they run in-house, the choice of how to proceed with the use of cloud technology is not so clear. They can consider the choices of creating private clouds or using hybrid clouds. Private Clouds Many industry analysts believe that private clouds will be attractive to in-house IT groups for the foreseeable future because of concerns about governance, data security, and performance management.[8] Private clouds also offer large companies an inviting way to consolidate data centers, cut technical support and operations staff, and increase server utilization. Typical server utilization inside corporate data centers ranges from as low as 2 percent up to around 10 percent, and implementing a cloud can raise those levels to 60 to 70 percent,[9] saving the need to buy a lot of additional servers. There is a case to be made that private clouds don't need to be quite as automated and self-served as public clouds to deliver value to companies in the form of increased server utilization and faster user provisioning. Instead of using online Web request forms to provision computing services for a new application system, employees of a company could just send an email to their IT provisioning group with the request. The in-house IT organization could get it done and email the requestor back in a few hours with the confirmation and information they need to start using the newly provisioned system. There is also a case to be made that larger companies enjoy certain economies of scale in the IT operations, and they can provide IaaS and PaaS less expensively than those services could be obtained from outside cloud service providers. In this circumstance, private clouds make good business sense for certain categories of services. Private clouds may not need to run entirely on uniform hardware the way public clouds do. For instance, IBM has experience building private clouds that use products such as Tivoli on its mainframes, Windows and Linux on its servers, and Websphere transaction management and SOA and MQ Series for message sharing between these different platforms. By doing this they are able to create fit-for-purpose clouds and increase the utilization of each platform. Building a private cloud in the typical heterogeneous enterprise environment offers advantages such as: • • • It enables IT organizations to leverage existing infrastructure and get cost-effective use of their previous IT investments. Placing cloud computing inside the corporate data center eliminates many of the issues that accompany the use of public clouds such as data security, performance management and SLAs, and concerns about regulatory compliance. Private clouds also have the potential to obtain a lower cost of use, since they don't have a profit margin added onto their services, as is the case with public clouds. In building their private clouds, companies can take an approach that allows them to invest in their private cloud as the first step on a journey to become comfortable with a cloud operating model. Companies may feel that external cloud environments have too many unknowns and too much risk until they become accustomed to this new operation model. Private clouds are a good way to test maturity and reliability of the technology. Companies can develop trust in the technology and the pubic cloud 17 | Chapter 3: Cloud Computing and the New Economics of Business providers they work with on a limited scale. They can learn to deal with different regulatory, data control, and security issues. Over time, the in-house private cloud versus public cloud mix goes from 9010 to something like a 50-50 mix or even a 20-80 mix. Large companies will probably not get to an all or nothing situation regarding their use of public and/or private clouds anytime soon. Hybrid Clouds A company may have a private cloud to share IT resources across multiple applications and increase utilization of the servers in their data center. Suppose that a company starts to experience a surge in user demand for one of their applications. By using a hybrid cloud they can quickly and cost effectively expand the capacity of the servers in their private cloud. They can draw upon the power of a public cloud to handle the increased user demand and maintain good system service levels for the people using it. To create a hybrid cloud, companies need to put the infrastructure in place that will allow them to integrate public clouds with their private clouds while still maintaining security and performance management capabilities. IT vendor companies (Cisco, Itricity, Juniper Networks, and Nimsoft to name a few) are making the technology that allows companies to do this. This is the underlying infrastructure companies need for hybrid clouds. Integration of cloud applications and in-house systems requires an effective method for maintaining security, monitoring performance, and passing data back and forth between the cloud and the in-house systems. Perimeter security in hybrid clouds can be provided by a number of methods, such as data encryption and virtual private networks. Many in-house IT groups are already familiar with the use of this technology. Most cloud vendors provide robust tools to help manage system performance in the cloud. This is their business, so they invest heavily in performance monitoring and reporting capabilities that often are superior to what companies have in-house. Once in place, this infrastructure becomes the base for new business models. It allows rapid expansion and contraction of computing power as business needs change, and it also provides the security, performance management, and regulatory compliance needed to operate hybrid clouds. Cloud applications are, by their nature, relatively easy to integrate with other systems because they are built with well-defined application interfaces, known as application programming interfaces (APIs). Compared with the task of integrating different in-house applications, integrating cloud applications with in-house applications is often easier because the APIs of cloud applications make it easier to import and export data and pass that data back and forth between the cloud and in-house systems. ISSUES TO CONSIDER WHEN THINKING ABOUT PRIVATE CLOUDS Private clouds have all the issues associated with public clouds, and they require significant upfront capital investment to set up. They have the same problems with monitoring and managing performance. They have the same problems with the risk of vendor lock-in (particularly with respect to the virtualization technology used)[10] and the question of whether that vendor will keep up with the pace of technology change in the marketplace. It is a huge challenge for internal IT organizations attempting to build private clouds because they often have not dealt with the business-process reengineering aspects of private clouds. The information 18 | Chapter 3: Cloud Computing and the New Economics of Business technology infrastructure library (ITIL) is a popular set of best practices that are widely used to run the in-house data centers of individual companies. ITIL practices are going to clash with cloud practices because ITIL is very manually intensive, and clouds, by definition, must be highly automated in order to achieve the levels of user self-service and the rapid infrastructure provisioning required for meeting user service requests. At present there are few, if any, in-house IT organizations that can match the operating discipline, automation, and resultant efficiencies of the big cloud data centers being run by vendors such as Amazon, Google, IBM, Microsoft, and Rackspace. Cloud vendors invest in their infrastructure and in automated systems administration capabilities, to achieve great economies of scale and operating efficiency. In-house IT groups are always being squeezed to save money and to cut their operations budgets, so they are challenged to create the economies of scale that public cloud vendors can achieve. As Irving Wladawsky-Berger put it, most company data centers are a hodge-podge of different technology reflecting the company's history with different vendors and its mergers and acquisitions. He pointed out that they look like what most factories looked like before the advent of lean manufacturing practices. The engineering disciplines promoted by lean manufacturing and implemented by the Japanese and Germans have set a standard that every other manufacturing company needs to match if they want to achieve world-class productivity and cost efficiency levels. Companies need to adopt similar practices regarding their equipment and the layout and operations of their data centers. Public cloud vendors are bringing this same discipline to bear on their cloud data centers. The public cloud vendors have implemented a new world-class level of practices and use of equipment that inhouse IT groups must also adopt if they wish to achieve the same level of productivity and efficiency. Based on his experience over the last couple of years, senior enterprise architect Rick Pittard put it this way: Size of data centers is important but may not be as central to gaining economies of scale as having a standardized hardware and operating system environment. Size and standardization are both necessary to get real economies of scale. If you have size, it can reduce cost if you also have standardization, but without standardization, size alone will not reduce your costs. Without standards, the diversity of hardware makes it very hard to move into a cloud environment.[11] Cloud computing by its nature requires a lot of innovation; it demands steady innovation to make it work and to make it easy to use by a mass consumer market. Public cloud vendors in the business-toconsumer world have innovated rapidly and done so over the last several years, in spite of our difficult economic situation. Where in the corporate world has so much changed in so short a time? The innovation cycles of public cloud vendors are usually much shorter than most corporate IT lifecycles; most companies work on five- to six-year lifecycles or longer. It will be a challenge for them to keep up with the pace of change initiated by public cloud vendors. The rapid innovation cycles of public cloud providers like Google and Amazon are driven by real-time customer feedback loops. As discussed in Chapter 1, customer feedback drives their innovation in a much more effective manner than the feedback that drives traditional IT vendors and in-house IT groups. That is because the central business of cloud service providers is to make money by responding quickly to customer desires. In-house IT groups do not have that stimulus, and they are not seen as 19 | Chapter 3: Cloud Computing and the New Economics of Business profit centers in their companies, so they do not have access to the same levels of investment to improve their service offerings. If not used carefully, private clouds can defeat the central purpose and the value proposition that cloud computing provides to companies. With private clouds, companies still have the distraction of buying servers, building data centers, and operating enterprise systems. In-house IT staff is still focused on running existing technology and systems instead of figuring out what new technology and systems the company needs. THE CLOUD IS A PLATFORM FOR MANAGING BUSINESS PROCESSES In this real-time global economy with short product lifecycles, companies need to be good at bringing new products to market, and they need to be good at tailoring existing products to keep up with shifting customer preferences. In their ongoing search for new products and new markets, companies engage in growth by expansion of existing business units and by mergers and acquisitions of other companies. Business processes need to be flexible to accommodate these activities. The value proposition delivered by most companies (unless they are themselves cloud service providers) lies in the way they manage their business processes, not their technology. It lies in the way they tailor their processes to meet constantly changing market conditions. Technology is a means to this end, but technology is not an end in itself. Although the cloud is certainly a platform for managing the delivery of computing services, that perspective is more from the traditional technology-oriented view. Another way to look at the cloud is from the business perspective of companies that use the cloud to support their operations. From their perspective, the greatest benefit they can gain comes not from cost savings in technology, but from the revenue they earn by being more responsive to changing customer desires; the revenues they generate with faster roll out of new products and successful expansion into new markets. In addition, it is important to remember that companies are much less self-contained and much less vertically integrated than they were 20 years ago. Companies have been steadily outsourcing non-core activities so that they can concentrate their time and money on doing the value-added things that produce the product or service their customers want. As a consequence, most companies are dependent on a network of suppliers that provide support services. And for companies to manage their business processes effectively, they need to find ways to effectively collaborate with their supplier partners. Exhibit 3.5 illustrates the interconnectedness that now is the norm for most companies. Exhibit 3.5: Interconnected Enterprises The cloud is not about new technology so much as it is about new business models. The business model in Exhibit 3.5 shows how 20 | Chapter 3: Cloud Computing and the New Economics of Business companies are evolving away from self-contained organizations that perform all their core and support activities internally. Companies are becoming enmeshed in networks of suppliers and customers, and to paraphrase a famous saying, no company is an island. No company can succeed all by itself. Companies depend more than ever on effective collaboration with its supplier partners. Business services need to be delivered in a reliable and predictable fashion, and it is the flow of information back and forth between companies via the cloud that makes this possible. The cloud is an ideal environment for companies to build and deliver an inventory of business services targeted to individual market segments and specific customers. Business process management (BPM) is the foundation for offering business services over the cloud. Using BPM, companies working together can assemble an appropriate bundle of business services needed to best serve certain kinds of customers. BPM systems can then monitor performance of these inter-company services and provide all parties with the real-time reporting and transparency they need, so as to continually adjust the business processes that deliver these services as conditions change. BPM systems tap into and monitor the data that flows between the transaction processing systems used by companies. For instance, consider a business process like that which receives orders, routes the orders for fulfillment, ships the ordered products, and then bills the customer. There are several transaction processing systems involved: a Web-based product catalog and order entry system; an order routing system; an order fulfillment system; a billing system; and an accounts receivable system. You can think of the workings of each of these individual systems, but to really understand what is happening and optimize the whole process of taking orders and serving the customer, you need to see and understand how all these systems work together. A BPM system can show you a unified big picture view of the data flowing between these systems, and it can spot bottlenecks where data flow slows down or where problems develop that affect the efficiency of the whole process. A recent report from Gartner states, "By 2013 dynamic BPM will be an imperative for companies seeking process efficiencies in increasingly chaotic environments."[12] It also says that more and more customerfacing processes will be configured based on specific knowledge about the customer, and suppliers will use BPM to tailor their processes on a just-in-time basis to meet the evolving needs of customers. In other words, BPM systems deployed in a cloud environment become the way that companies working together in value chains deliver a constantly changing mix of responsive products and services to customers. Cloud-based BPM becomes the basis for effective cross-company collaboration to deliver personalized and specialized processes that support individual customers. BPM and SOA are two sides of the same coin: BPM is the business view; SOA is the technology view. BPM allows companies to model their business processes and combine and streamline them. SOA allows companies to reuse software assets and cost-effectively create systems to support redesigned business process flows. Companies can use SOA or WOA to integrate across different clouds and integrate cloud applications with internal company systems. BPM allows companies to break their business processes into collections of interconnected tasks. This is an important step in enabling them to extend their operations beyond their own company boundaries, to embrace services provided over the cloud. In this way, they can outsource certain processes and tasks so that they can concentrate on their core value added processes and continue to improve them and invent new ones. 21 | Chapter 3: Cloud Computing and the New Economics of Business In a world where products and services quickly become commodities and where profit margins are constantly being squeezed by this commoditization, it is the ability to continuously tailor these products and services that will earn companies an additional profit margin on top of the diminishing margins offered by those otherwise commodity products. The driving force of the responsive economy of this century is coming from the unleashing of innovation via cloud services that are quickly becoming available all over the world. Scientists can collaborate on health and environmental problems; businesses can cooperate to deliver tailored products and services worldwide; and in these business networks, companies can focus on doing what they do best and rely on other companies to perform the complementary tasks that are a part of delivering the finished product to the end-use customer. It is through this agile process of continuous response to changing customer desires that most companies will differentiate themselves and find most of their profit opportunities. You could call the profits generated in this way as the "agility dividend." Real-time global markets are continuously adjusting the price of commodity products towards their cost of production, just as real-time stock markets continually adjust the price of stocks. So it is in the agility dividend that most companies will find their best opportunities to generate profitable revenue that is above the prices set by global markets. AUTOMATE ROUTINE PROCESSES, FOCUS PEOPLE ON HANDLING EXCEPTIONS Companies are like fish swimming in oceans of data. There is simply not enough time for people to handle or review standard information that records just the predictable, expected operations occurring in a company. Routine, standard data and procedures for handling it must be driven by automated transaction systems that support the company's standard operating procedures. Computers handle routine situations much better than humans because they never get bored by the routine, and they scale up quickly as transaction volumes increase. People in a responsive organization need to devote their time to handling nonstandard data. Nonstandard data is any kind of data that is different from what is expected: data that for any reason does not conform to the rules built into automated transaction systems or the performance parameters built into performance monitoring systems. When a company's systems encounter this kind of data, people quickly get involved. The greatest opportunities for any organization lie in the way they detect and respond to unexpected problems, threats, or opportunities. Use computers to do what they do best. Let them handle the day-in, day-out moving of routine data related to basic transactions such as purchase orders, invoices, account balances, order status, address changes, and so on. Wherever there are people doing routine data entry or repetitious work of any sort, this is an opportunity to automate. Computers do this sort of work much better, faster, and cheaper. Use people to do what they do best. What they do best is think, communicate, and solve problems. We don't need to build excessive amounts of complexity and cost into new computer systems if we free up people from routine work and give them the data and training they need to solve complex problems and handle the exceptions to routine operations. We don't need artificial intelligence in our systems when we can apply the real intelligence of people who are trained, motivated, and empowered. Most business operations are routine, and repetitive work that can be handled with relatively simple sets of processing rules can be applied to business processes with BPM systems. The BPM systems can 22 | Chapter 3: Cloud Computing and the New Economics of Business be used to move and monitor data between different systems and different companies. They monitor the processes they support, and when they detect a problem or a slow-down, they send alerts to people assigned to handle those issues. Whenever a transaction happens that does not follow one of the simple routine processing rules, the BPM system traps the data related to that transaction and notifies a qualified person. People will either be able to correct the data so that it fits back into a simple, predefined process, or they will take care of the transactions themselves all the way through. They will have time to do this because they won't be bogged down and worn out doing the routine stuff. This is where they can generate the most value for customers and for the company; this is known as generating agility dividends; and people will do a great job on these tasks, too. Since they are non-routine, they are interesting. They involve thinking, communicating with others, and problem solving. People like doing this kind of work. It's fun. The human brain is more fine-tuned than any computer to do just this kind of work. By automating the mass of rote, routine, and repetitious work, your organization will get great cost efficiencies. By empowering people to handle the non-routine tasks, companies will become very responsive to unique customer needs. It is this blend of efficiency and responsiveness that enables a company to outperform its competition. Cost savings happen when companies are able to act in a coordinated manner to continually optimize their common and individual business processes and adjust to changing business conditions on a daily basis. These continuous adjustments to changing conditions result in a stream of cost savings generated by optimizing operations to adjust to changes in prices of labor and raw materials and changes in supply and demand for a company's products and services. Profits are generated by continuous tailoring of products and services to changing customer needs and desires. Products that are tailored to meet demanding customer service requirements are worth more than the commodity versions of those products. This tailoring of products results in sales revenues that are slightly higher than the market average, because customers are willing to pay a few percent more to get those tailored products. FOUR TECHNOLOGIES ENABLE RESPONSIVE BUSINESS PROCESSES There are four application technologies that companies can use in various combinations and become very responsive to changing conditions and emerging threats and opportunities: 1. 2. 3. 4. Business process management (BPM) Complex event processing (CEP) Business intelligence (BI) Simulation modeling BPM is a way for companies to observe productivity in their operations and carry out a continuous, incremental process of improving operational performance. A company starts by mapping out its key processes and defines the steps or work tasks in each process. Then it uses BPM software to collect and display a continuous stream of data that shows the movement of transactions through each step. The BPM software can be used to automate many of the routine tasks, such as moving different kinds of data from one task to another. It can also be set to detect certain error conditions and send automatic alerts to people who need to respond to these conditions quickly. 23 | Chapter 3: Cloud Computing and the New Economics of Business CEP complements the capabilities of BPM systems. A CEP system can monitor multiple data streams and can do real-time comparisons of data in these data streams to detect predefined patterns in the data that indicate the occurrence of certain events. Companies can define specific event patterns that then trigger the system to make certain responses. Some events may trigger the system to set in motion a set of online calculations and responses that react to those events. Other events may trigger the system to send alerts to people. For instance, a CEP system used by an airline might detect that an airplane is at a departure airport and that a large storm is heading toward that airport. It also detects that a boarding gate at the arrival airport is reserved for that airplane. In these data streams, the system can detect that the airplane's departure will be delayed by the approaching storm and that the time slot reserved for the boarding gate at the arrival airport has to be changed. The system can then make the change itself or it can send an alert to airline staff so they can respond appropriately. BI systems collect, store, and analyze data. These systems allow people to orient themselves and decide on what actions to take. They collect data from many different sources. Data can be collected from sensors and radio-frequency identification (RFID) devices. Data can be collected by BPM systems or data can be obtained from the many transaction processing systems in a company such as ERP systems, order entry systems, or CRM systems. Once the data is collected, it is stored in a database where people access it as needed. Often the database is updated with new data on a continuous or real-time basis, and summary displays of relevant data are available to people through Web-based dashboards. When people access the data, they use BI software tools that help them analyze the information and display the results. BI software tools run the gamut from simple spreadsheets and charts to complex multivariable regression analysis and linear programming. The proper mix of BI tools is determined by the needs of the people in a situation and their skill and training levels. The combination of BPM and CEP and BI systems is sometimes referred to as enterprise performance management (EPM). Simulation modeling software is a category of software that is starting to get increasing attention. Because of the fast pace of change in business, companies are faced with the need to make important decisions more often, and these decisions have significant consequences on company operations and profitability. Companies increasingly need to make decisions about how to best operate in conditions they have not encountered before. Simulation modeling has powerful capabilities for improving the quality of decision making in an organization. Simulation modeling software allows people to create models of things such as a factory or a supply chain network or a vehicle delivery route. Then they can subject the models to different inputs and different situations and observe what happens. A design that may seem good on paper could very well turn out to have problems that are not apparent until the design is modeled and its performance is simulated under a range of different conditions. It is much faster and cheaper to discover problems through simulations than to find out the hard way through real experience. Existing transaction systems, such as ERP, order management, accounting, inventory management, delivery scheduling, factory control, and maintenance systems provide a steady stream of data that reflects individual processes in a company or between groups of companies. This data can be monitored through the use of BPM and CEP systems to provide a comprehensive end-to-end picture of the productivity and performance levels in these operating processes. BPM systems can update this picture on a real-time or nearly real-time basis, and show people where the bottlenecks and disruptions are that need their attention. 24 | Chapter 3: Cloud Computing and the New Economics of Business Once people have identified the bottlenecks and disruptions, they can make use of BI databases and analytical software to investigate the problems and identify their root causes. When root causes are identified, people can design ways to address them. Then, by using simulation systems, they can model potential process changes and see the probable impact of each different change. In this way, people select the most effective changes and implement them with a high level of confidence that they will actually deliver the desired results. The power of these four technologies is multiplied when they are used together and when they are used on a universally accessible platform such as the cloud. This is because they enable more effective and timely collaboration among companies working together. When these technologies run in the cloud, people in all participating companies can see data and status of operating processes in real-time, and this transparency enables effective and timely brainstorming and problem solving. Companies have the opportunity to collaborate and design extraordinarily responsive business processes to drive their operations, enabling them to continuously adjust their operations to changing and unforeseen developments in information technology. NOTES 1. Joseph A. Schumpeter, Capitalism, Socialism and Democracy (New York: Harper, 1975) [orig. pub. 1942], pp. 82–85. 2. Frank Gens, "Defining ‘Cloud Services’—An IDC update," IDC Exchange, (September 30th 2009), http://blogs.idc.com/ie/?p=422. 3. Darryl Plummer, "Experts Define Cloud Computing: Can We Get a Little Definition in our Definitions?" Gartner article, 2009, http://blogs.gartner.com/daryl_plummer/2009/01/27/experts-define-cloud-computing-can-weget-a-little-definition-in-our-definitions/. 4. Jeff Kaplan, "Simplifying the Term ‘Cloud Computing’" Datamation.com blog, (June 25, 2009), http://itmanagement.earthweb.com/netsys/article.php/3826921/Simplifying-the-Term-CloudComputing.htm. 5. Frank Enfanto, VP of Worldwide IT Operations, ACI Worldwide, private email correspondence with author (December 22, 2009). 6. Rick Pittard, telephone interview by Michael Hugos, February 10, 2010. 7. Ibid. 8. Bechtel Corporation, "Pioneers of the Private Cloud," December 21, 2009, www.alacrastore.com/storecontent/tectrends/177031. 9. Ibid. 10. Bernard Golden, articles on private clouds at CIO.com, 2009, www.cio.com.au/tag/Bernard%20Golden. 11. See note 6. 12. Gartner report, January 13, 2010, www.gartner.com/it/page.jsp?id=1278415. 25 | Chapter 3: Cloud Computing and the New Economics of Business JOURNAL OF NETWORKS, VOL. 7, NO. 2, FEBRUARY 2012 305 A Review on Cloud Computing Development Sugang Ma South China University of Technology, Guangzhou, China Email: crmsgo@126.com Abstract—Cloud computing can provide critical services for business management, reducing IT costs and maintenance costs of hardware and software effectively. In the meanwhile, it can enable enterprises to access to professional IT solutions with less IT investment. Cloud computing is of great significance for the ICT industry of each country. It is now bringing enormous impact to the human society, especially the business world. In this paper, the basic concepts and the development of cloud computing were introduced, and then the current situation and development of cloud computing research from two aspects of technology and business were illustrated. Finally future trends of cloud computing were discussed. Index Terms—cloud computing, IBM clouds, Google clouds, Amazon clouds I. INTRODUCTION Cloud computing refers to the logical computational resources (data, software) accessible via a computer network, rather than from a local computer. It refers to both the applications delivered as services over the Internet and the hardware and systems software in the datacenters that provide those services. The services themselves have long been referred to as Software as a Service (SaaS), so we use that term. The datacenter hardware and software is what we will call a Cloud [2]. Cloud computing is a concept put forward by Google, which is a new way to play with computer and Internet. It is a kind of computing which is based on the internet. Shared information is provided to computers and other devices on demand, like the electricity grid. As a beautiful web application model, it is a paradigm following the transferring from mainframe to clientserver in the early 1980s.The technology infrastructure “in the cloud” supports details without users controlling over. Cloud computing describes a new consumption, supplement, and delivery model for IT services based on the Internet. It usually involves over-Internet dynamic scalability and virtualization resources are often provided. This is a by-product and easy access to remote computers via the Internet web site results. It aims at relatively low cost of the network to the calculation of multiple entities into one powerful computing capability with the perfect system, and use of SaaS, PaaS, IaaS, MSP business models and other advanced computing power of this powerful distribution to terminal users’ hands. The word "cloud" is used as a metaphor for the Internet, used to represent mobile phone networks on the basis of the past, and later to depict itself as an underlying infrastructure that represents an abstract computer network mapping of the Internet cloud. Cloud Computing is a core concept of © 2012 ACADEMY PUBLISHER doi:10.4304/jnw.7.2.305-310 continuous improvement through the "cloud" of processing power, thereby reducing the processing burden on the user terminal, which eventually simplified into a simple input and output devices, simultaneously enjoying the "cloud" computing power of ability when needed. A typical cloud computing provider provides a common business online application which is accessed from another Web service or Web browser as software, and software and data are stored on the server. A paramount element of cloud computing is customization and the creation of a user-defined experience. Many experts are researching cloud computing. In July 2008, HP, Inter Corporation and Yahoo! announced the creation of a global multi-data center, open source test bed, called Open Cirrus ,designed to encourage research into all aspects of cloud computing. In July 2010, HP Labs India announced a new cloud –based technology designed to simplify taking content and making it mobileenabled. Site on Mobile is designed for emerging markets where people are more likely to surf the net via cell phones than computers. The IEEE Technical Committee on Services Computing in IEEE Computer Society sponsors the IEEE International Conference on cloud computing (CLOUD). CLOUD 2010 was held on 2010 in Miami, Florida [6]. Years ago, grid computing became the hot issue in the IT world for it opens a new era of integrating resources to solve problems. The same idea (distributed computing) is shared with cloud computing. However, cloud computing is considered to be the technology derived from grid computing with different problem-dealing-aspects of processing but core concept unchanged. The paramount notion of grid computing is collecting all available resources to solve problems (mainly scientific ones) that individual computing center cannot deal with. The most famous case of grid computing, SETI@HOME, searches for extra-terrestrial intelligence using the spare resources from the volunteer Internet-connected computing computers all over the world. On the other hand, cloud computing aims to supply efficient, qualified services with part of the available large scale of computing resources [5]. In this paper, we first introduce the basic concepts and the development of cloud computing. The second part is about key technology of cloud computing, including data storage, data management and programming model. Thirdly, the article will try to explain the impact that cloud computing brings to the business world by offering the examples of three widely accepted cloud services. The next part is about the future challenges of cloud computing. Summary is concluded in the last section. 306 JOURNAL OF NETWORKS, VOL. 7, NO. 2, FEBRUARY 2012 II. KEY TECHNOLOGY OF CLOUD COMPUTING Cloud computing provides the most reliable and secure data storage center. Users no longer have to worry about data loss, virus attack and other problems. Because the world's most professional team are helping you manage information, and the world's most advanced data center is helping you save the data. The cloud computing has the minimum requirement on the client devices, which makes it useful and most convenient. In addition, you can easily realize sharing data and application between different devices. For network applications in the cloud model, data is only one, saved in the "cloud", on the other side. You only need all the electronic equipment connected to the Internet to simultaneously access and use the same data. Cloud computing provides us almost infinite possibility using the Internet. Personal computer or other electronic devices can not provide unlimited storage space and computing power, but in the "cloud" the other side, by the thousands, tens of thousands or even more servers composed of large clusters this can easily be done. Personal and individual devices is limited, but the potential of cloud computing is nearly limitless. In this part, we will introduce key technology of cloud computing, including data storage, data management and programming model. A. Data storage Cloud computing stores data in large scale distributed systems in order to provide the most reliable and secure data storage center. Users no longer have to worry about data loss, virus attack and other problems. Because the world's most professional team are helping you manage information, and the world's most advanced data center is helping you save the data. Cloud computing needs massive data storage, but also needs to meet high availability, high reliability and economy, etc. As a result, systems in cloud computing needs to support large data sets, process mode of write once and read many, and have high concurrency. One of the most important data storage technology of cloud computing is Google File System (GFS). It is constituted by a Master and a large block of servers. Master stores all the Meta data of file system including namespace, access controlling file block information, the file block location information etc. GFS file are cut into 64MB blocks for storage. GFS file system uses a redundant storage means in order to ensure reliability of data. Each data is saved by more than 3 copies in the system, including two copies in different nodes of the same rack in order to take full advantage of the bandwidth within the cabinet while the other copy is stored in different nodes of the rack. After obtaining the location information of the target data block from the Master, the client does not read data through the Master but interacts the block server directly. B. Data management Cloud computing system has to process and analysis large data sets as to provide high efficient service. Thus © 2012 ACADEMY PUBLISHER data management technology should manage large data sets with high efficiency. What’s more, how to find specific data in large data sets is also a key problem in data management [3]. The famous data management technology is the large-scale database system Big Table with the weak consistency requirements [4]. Data items in Big Table are arranged according to the order of line keyword in the dictionary. Each row is assigned dynamically into record tablet. Each tablet server node is responsible to manage for about 100 record tablets. Time stamp is an integer of 64 bit, which indicates different versions of the data. Column family is a collection of several columns, Big Table access permissions is controlled in family size for the column. Big Table system depends on the underlying structure of the cluster system, which includes a distributed cluster task scheduler; the GFS file system which has been addressed above, and a distributed lock service Chubby. Big Table use Chubby which is a very robust coarsegrained lock to save the pointer of Root Tablet, and use one server as master server to conserve and operate metadata. When a client reads data, first Root Tablet location information and meta-data table Metadata Tablet location information are obtained from the Chubby Server, second the User Table of location information including the target data location information are read from the Metadata Tablet, and then the target data location information item are read from the User Table. Big Table’s main server not only manages the metadata but also is responsible to remote manage remotely and allocate for the Tablet Serve; Client-side proceeds control communication with the main server to obtain metadata through the programming interface, and proceeds data communication with the Tablet Server that is responsible to deal with specific read and writing requests[1]. C. Programming model In order to make customer enjoy cloud computing service more conveniently, programming model in cloud computing must be very simple. Nowadays, most cloud computing systems Map Reduce programming model. Map Reduce programming framework is to support parallel computing. Map Reduce is not only the programming model that processes and brings about a large data set but also an efficient task scheduling model. It is through two simple concepts about the "Map (map)" and "Reduce (simplified)," that constitute the operation basic unit. The programmer can finish the distributed parallel program development who specifies the data processing to each block data in the Map function and how to regress the intermediate results of the block data processing in the Reduce function. When Map-Reduce program is running in the cluster, programmers do not care how to block, allocate and schedule the input data; even more the system will manage the node failure in clusters and communication between nodes. The implementation of a Map Reduce program requires five steps: inputting file, one file being assigned to many workers parallel, writing intermediate file (local writing), many Reduce workers running, outputting the final result JOURNAL OF NETWORKS, VOL. 7, NO. 2, FEBRUARY 2012 [1]. Writing intermediate file locally decreases the pressure on the network bandwidth and saves the time on writing intermediate file. If successful, it reads the data needed from the node where the intermediate file is with remote procedure call according to the location information of the intermediate file obtained from the Master. Public or external cloud introduces the traditional mainstream consciousness in which resource is dynamically fine-grained, on the basis of self-service provisioning via the Internet cloud applications through the network (network service), from off-site third-party providers who bill fine-grained basis for utility computing. The idea of a Private Computer Utility was first illustrated by Douglas Parkhill in his 1966 book "The Challenge of the Computer Utility". The idea was based on direct comparison with other industries (e.g. the electricity industry) and the extensive use of hybrid supply models to counterbalance and mitigate risks. Private cloud and internal cloud have been described as neologisms, despite that the concepts themselves pre-date the term cloud by 40 years. Even within modern utility industries, hybrid models still exist despite the formation of reasonably well functioning markets and the ability to combine multiple providers. Some vendors use the term to describe the product ...