For more than four decades, the information management needs of the enterprise have been most often met by the construction of datacentres that are owned and operated by the individual organisations requiring that capability. In the same way that a manufacturing site might once have had a water wheel as its source of motive power, an enterprise datacentre had its mainframe computer or perhaps a cluster of variously sized computers for different tasks.
Electrical power is now more economically generated and delivered through regional systems that employ shared generator sites: these are constructed and operated at far lower cost per unit of capacity than smaller generators for individual factories. As noted by Nicholas Carr, a similar transition is now observed in the delivery of functions that store and manipulate data for a wide range of business functions.
Rapid growth of computer speed has been accompanied by rapid decline of computer hardware cost. This combination of changes often leads to a belief that the basic economics of computing are continually improving, so that there is no need for fundamental change in the manner of constructing and delivering the functions of information management.
Closer inspection reveals that there are fundamental barriers to continuing along the path of technology and practice that has defined the development of business information systems since the late 1950s. Further, there are important new developments that create more attractive alternatives.
The letter of Moore’s Law
Often repeated is the statement that computers become twice as powerful, at any given price, every 18 months. This trend is often mislabelled as “Moore’s Law,” making reference to a prediction stated in 1965 by Gordon Moore.
It is notable that Moore, who was one of the co-founders of Intel actually said something significantly different. Moore predicted that the technology of producing integrated circuits would develop in such a way that the design of lowest cost would package twice as many transistors into the same physical space, in a cycle that would continue to repeat at a constant rate of exponential growth. For several generations of technology, this growing complexity of circuits did relate directly to growing performance: in particular, from the mid-1970s through the late 1990s, cost and performance of personal computers reliably improved in this manner.
As the 21st century began, continued growth in the complexity of microprocessors did not continue to bring forth the previous rates of improvement in performance of common tasks. Designers of electronic components have found it more effective to construct computer processors with several identical processor units, commonly known as “multi-core” processors.
One can quickly understand that the most direct way to apply a multi-core processor is by sharing its power in a multi-user environment, assigning tasks in a simple way to individual cores. The growing speed and the global reach of the Internet arrive at the proper time to permit this sharing in a convenient and economical manner.
Moore’s Law, one may therefore say, has not been repealed, but its jurisdiction has changed. The frontier of continued improvement in price and performance of computer functions will no longer be on the individual desk or laptop machine, but in the shared datacentres that perform computing and collaboration tasks for global communities.
Design for global scale
One does not achieve global capability by taking the software design of the past and doing the same thing at a larger size. Previous software designs contain, and reflect, the assumption that only a single person or a single commercial enterprise are being served. This assumption leads to certain results: for example, it has long been assumed that the function of a packaged software product will be adjusted to meet the needs of a single customer by making changes to the software code. This manner of adjusting software function cannot be accepted if the software product is operating in a shared environment, since different modifications are needed to serve the individual requirements of different enterprises in different industries.
When a software product is built from the outset for use in a widely shared manner, the resulting conception and construction lead to fundamental changes in software design. To state the principle in a simple form: all aspects of the software operation that will be common to all users should be in a layer of software “code” that is shared, while all aspects of the software operation that are individual to each user should be defined in a higher layer of specifications called “metadata.” The notation of “metadata” represents the knowledge and the business practices of each separate organisation that relies on the shared system’s functions.
When all of the customisations of the users are placed above the metadata boundary, the software that performs shared functions can be improved without disrupting user operations. In the previous technology environment of the 20th-century datacentre, an “upgrade” was a process that consumed much time and presented great risks of complication and cost. These painful problems persist today in many organisations. In the 21st-century environment of the shared system with metadata customisations, upgrades to shared capability are performed with no need for action by users. It is vital to understand the additional benefit that there is also no need for users to repeat their past efforts, reworking them for compatibility with updated foundations, to enable strategic business process improvements. The metadata model preserves the business value of previous work even as underlying foundations are dramatically improved.
The promise of cloud computing
It has been established above that fundamental trends of technology and necessities of economics are aligned in the direction of delivering information systems capability as a service. It has further been argued that the crucial principles of multi-tenant architecture and metadata-based customisation are principal factors in achieving the full potential of the service delivery model. Also explored have been the common concerns for ease of information asset integration and protection of information asset security, both of which are readily achieved by proper design and operational practice of a Platform as a Service.
It now becomes "page1" possible to look at the Platform as a Service as more than an economical successor to 20th-century models of mainframe or client-server technology. Platform as a Service is not merely competitive against what has come before: it is a compelling opportunity to enable new modes of engagement with customers and collaboration with business partners.
Focus on function and value
Research by the analysis personnel of Gartner has confirmed what most computing practitioners already suspect. Most of the time and money that are devoted to information technology efforts in large organisations are consumed in continuing with current capability, leaving only a small fraction of those resources for the creation of new value.
Adoption of Platform as a Service takes entire categories of capital investment and operational expense off the agenda of the enterprise IT team. Creation of a new business function no longer begins with a lengthy, cumbersome process of constructing new facilities and installing local copies of packaged software. The operating capacity is available on demand across the Internet; the common functions of database management, electronic mail integration, data backup and disaster recovery, and other such tasks are provided on an ongoing basis by the service provider. High levels of security are already in place, with no need to acquire and configure new systems.
The resulting benefits are demonstrated in actual projects. CODA, a United Kingdom software provider with 30 years’ experience in older application development models, made the decision in 2007 to bring its next generation of financial software capability to market as a Force.com Platform as a Service application. The CEO of the company, Jeremy Roche, has made public statements that this decision “saved at least two years in elapsed time and at least 25 man-years of development time.”
Estimates by Gartner find that it is common for 90 parts in every 100 of the information systems appropriation to be what Gartner has called “dead money.” One may look at this number from another point of view: if the fraction of technology expenditure that goes to maintenance functions could be reduced to only 70 parts per 100, or even to as little as 30 or 20 parts, then the resources available for innovation would be multiplied by a factor of three, or seven or eight (respectively) without any increase in overall technology costs. This is the next chapter to be written in the book of Platform as a Service.
content management
Electronic mail has become a pervasive tool, but it is a poor choice for many tasks. When many individuals are involved in creating a single work product, often does one see many versions of a document being sent to members of that group to receive their suggestions and changes. Often is there confusion about the version of the document that is most recent.
Platform as a Service enables a new approach in which content collections are managed in a central location, and individuals have access to those collections through network links that always return the most current version of a work in progress. Also convenient in this arrangement are mechanisms for placing notations on a document to describe the situations in which it has been valuable. Previous versions of the document can also be made available without wasteful and confusing duplication of data on many separate users’ personal data storage devices. Further, the specific users who have examined a document and the dates and times of that access may readily be recorded.
Tools for collaboration
Much confusion and waste results when different individuals make independent efforts on shared tasks. The older Information Technology model, with software packages running separately on individual desktop and laptop computers, requires additional cost and complexity to create collaboration arrangements. The Platform as a Service model cures the disease, not the symptom. Rather than starting with separate, redundant tasks and then expending additional resources to bring them together, the Platform as a Service enables a single shared task environment in which many contributors can participate in a single concurrent effort.
This process of collaboration is now being widely enjoyed through the Google Docs service. Google’s system enables simultaneous editing and review of a document or a spreadsheet by many cooperating individuals, who require only shared permissions to participate and an Internet connection as the means of access. Strategic partnership between Google and salesforce.com enables automatic process performance and workflow that combines the functions of the Google productivity tools and the activity tracking capabilities of the salesforce.com applications.
Implications for technology practitioners
The transformation of the Information Technology discipline has many precedents. One often finds this change compared to the development of the power generation industry, as it was at the beginning of this article. One might also compare this to the manner in which telecommunications services are acquired by the enterprise, with telecommunications managers spending their time primarily on the negotiation and administration of service agreements rather than the installation and maintenance of on-premise hardware.
The efforts of the forward-looking information systems professional will be directed, increasingly, toward conversation and strategy formation with business unit managers and partners. Those conversations will be able to explore the manner in which the pervasive connections of the Internet and the efficient automations of the Platform as a Service can best be combined in creative ways to yield strategic business value.
The internal technology manager and staff are not being replaced by the external service provider. Rather, the time of the enterprise technology organisation is being made available for tasks that have much greater benefit to the enterprise, its partners, and its customers.
