The problems associated with the management of services in today's IT infrastructures do not exist in isolation. They are the result of the evolution of the services and the infrastructure they run on. More specifically, they are the result of the trend towards network-centric services and the continuing growth of network bandwidth, resulting in the data center becoming a fabric of resources, rather than a collection of discrete systems. These trends have resulted in the current service-oriented architectures and implementations with their benefits (performance, scaling, and potential resilience) and their problems (unmanageable complexity, lack of agility, fragility, and low utilization).
The same trends that resulted in the current implementations, and their attendant problems, also point toward the nature of the solution to those problems. The data center network, as a whole, is a new class of system. Thus, the problems are not those associated with managing a collection of systems. The problems are those of managing or building a new class of system. Rather than being connected to the network, this new system has the network at its core.
Accepting this assertion leads to the definition of the N1 Grid system. To build N1 Grid systems, their principles of operation must be established and systems architectures defined. These enable root-causes of today's IT infrastructure problems to be addressed. The N1 Grid vision defines the overall nature and attributes of an N1 Grid system. These systems span the data center network and include resources such as network switches, disk arrays, servers, and operating systems. They have workloads of high-level network centric services or applications that are managed through the automatic application of policies that are expressed as high-level business goals, ranging from quality of service to cost and utilization.
The N1 Grid solution strategy and architecture drive the evolution of existing products and the delivery of new products, as well as the evolution of today's infrastructure implementations, towards becoming N1 Grid systems. The consequences of this are discussed in the next chapter.