N1 Grid Systems Realized
Having explored some of the attributes and properties of N1 Grid systems and focused on the N1 Grid OE, this section explores what it means to realize an N1 Grid system. An N1 Grid system can be considered from two perspectives:
As a system, which is defined by its attributesthat is, composed of a set of network-distributed hardware and software components that are virtualized and that behave as a single system
This system is managed as a system, rather than as its discrete, individual components, and it is managed in a service-centric fashion.
As a system, which is defined by its implementationthat is, adhering to the N1 Grid principles of operation
This system is a realization of the N1 Grid architecture in which all of the architectural elements are realized as technology components that communicate with one another using standard IP protocols.
The second definition is more rigorous and describes a full implementation, yet there needs to be a means of evolving to this state with minimum disruption to existing data center infrastructure and practices.
Route to the N1 Grid Solutions
Infrastructure that realizes many of the basic attributes of a full N1 Grid system implementation can be constructed today by using both existing and custom components, and coordinated by using people and process. This is because most of the control and telemetry functionality required for the N1 Grid OE exists in a number of discrete products and tools.
Much of the value of an N1 Grid system is realized through its operational model (that is, the way in which it is managed). A user interface could be constructed that would be consistent with the full N1 Grid realization and that reflects the operational model. That interface could enable services and the service descriptions, together with the service-level objectives, to be defined and the underlying resource fabric to be described. All of this information could be stored in a traditional form, such as a database or spreadsheet.
The policies described could then be applied using people and process, where technology is unavailable. The process would be consistent with the N1 Grid principles of operation. When new technology is created that automates a new function, it could be integrated with the existing implementation.
This method means that a consistent management model can be instantiated. As new technology is rolled into the implementation, the management model need not change. All that would happen is that some tasks would disappear while the system managers would still interact through the user interface. Thus, there would be no need for them to acquire new skills every time a new implementation of an underlying mechanism is delivered. As technology is delivered, the overall abilities or attributes of the system improve as various tasks are automated (FIGURE 3-6). In fact, this could be considered virtualization of the data center management processes themselves or of the operational model implementation.
Figure 3-6 The Evolution of an N1 Grid System Implementation
Value of N1 Grid Systems
What value will fully implemented N1 Grid systems, as well as evolving N1 Grid implementations, actually deliver? The following lists the value:
TCO improvements through high-level policy-based management
This changes management scaling by hiding complexity so that rather than being based on numbers of servers, server vendors, number of operating system instances, versions and vendors, number of tools, and number of applications managed, it scales with the number and complexity of the services being offered.
TCO improvements through the application of a consistent operational model
The model reduces the need to continually invest in re-skilling operations staff in continually changing, upgraded product and tool implementations.
Improved efficiency and reliability through the formalization of many aspects of data center management
Currently, these aspects are ad hoc (for example, service life cycle management) and error prone. Ultimately, good practices and process become a part of the technology, in this case the N1 Grid OE.
Higher efficiency or resource utilization
Virtualization of individual servers, operating system instances, networked resources, and the automated mapping and remapping (provisioning) of workloads onto these resources, ensures higher utilization of resources.
Greater agility and responsiveness, from the definition and deployment of new services to changing resource allocations rapidly in response to business goals
Capacity planning becomes simpler because mistakes are more easily and dynamically corrected.
Greater reliability and availability through reduction of exposed complexity and automation of routine component-centric tasks
Although N1 Grid systems are intended to solve the most pressing problems within the data center, they do so in a manner that changes the perception of IT and that changes how services and the infrastructure they run on are paid for. Specifically, N1 Grid systems ultimately deliver transparency in terms of value delivered versus cost.
N1 Grid systems are driven by business goals, expressed as policies that cause workloads to be mapped and remapped as appropriate onto resources. To achieve this, the N1 Grid OE has to be able to reconcile resource consumption and the cost of those resources with the services that consume them. This is required to manage the quality of service goals, but it also means that N1 Grid systems should exhibit a previously unheard-of transparency with regards to reconciling the value delivered by a service and its components and the costs of providing that value. This type of transparency enables new business choices to be made, or at the very least, it enables existing business choices to be made with the right data.
For example, N1 Grid systems enable utility computing. Although utility computing is often referred to as a technology paradigm, it is probably best to think about it as a business paradigm associated with the delivery of services. Specifically, IT infrastructures, and the services that run on them, should be paid for in a way that is proportional to their delivered value. Thus, a service provider delivering a payroll service could charge per transaction or per employee on the payroll. They in turn could pay for their resources, both traditional hardware and metaresources, such as a database or an application server, based on usage. Traditional platform usage could be based on usage of certain resources, such as processors, memory, and storage. An application server could then be paid for based on transaction throughput and complexity.
The point is that after the telemetry to reconcile resource consumption with the services that consume them is available, there is sufficient data to enable more interesting and more flexible business relationships (for instance, relationships between system vendors and their customers, between service providers and their customers within large enterprises, and between data center staff and the business).
The potential exists for a new relationship between information technology providers and consumers. It would be interesting if when a CEO approaches the CIO and says "I'm afraid we need to cut your budget by 10 percent this year," the CIO could confidently say "Fine, however, you will lose this level of performance or capacity for your main customer-facing services. Marketing tells us that would translate into a 15 percent reduction in our customer base due to dissatisfaction." Indeed, a more interesting conversation, one which probably rarely occurs today, would be when the CIO says "If you invest x amount of money into our infrastructure, I can improve responsiveness and reduce outages by y, which means that we would over take our chief competitor" or "If you invest x, then we can bring new services to market 50 percent faster than our competitors."
Ultimately, N1 Grid systems, and those like them, will turn arcane technology infrastructures into genuine business systems, where the work they undertake is easily understood by the business and where much of the day-to-day management of the services they host is undertaken by line-of-business people, rather than IT specialists. Over time, it becomes the role of the IT specialist to install and deploy new services on N1 Grid systems, to describe them to N1 Grid systems so that the N1 Grid OE and the line of business can do the day-to-day management of the service, and to maintain the N1 Grid system.