Connections +
Feature

Focus on …Engineering & design: Evolving the “Next” Network

A combination of emerging elements is propelling us towards the "next" network. While a good deal of work will be required to achieve this goal, the labour will be amply rewarded.


April 1, 2002  


Print this page

T he communications business is in a time of change: a retrenchment from which a stronger, more viable, fundamentally altered network will emerge. This network will present greater scope, more efficient connectivity and utilization, more advanced functionality, more accurate metering and revenue generation, and greater potential than before.

Some predictions1 place the value of this potential at about US$20 trillion, with those same predictions indicating a significance curve that exceeds its tipping point in 2004.

PRESENT CIRCUMSTANCES

In today’s network, at any level of magnification, a tiered hub and spoke model prevails. This model requires that offerings within a tier be pitched at a similar level, with the perceived revenue opportunity declining with tier descent. In this circumstance, most offerings are pitched at the highest tier to the detriment of the model as a whole.

With this tendency to provide infrastructure and services at the highest tier, an unbalanced environment has emerged in which massive supply at the core is throttled at lower levels. This environment, embracing all elements of communications, is evident today: the emphasis on core development has left the rest of the network behind. Although some focus is now being seen in the edge environment, efforts exerted to date will only be rewarded when the latent demand for services at the access layer is unleashed. For providers of network infrastructure, capacity, services and colocation, a collective focus on the core has generated negative value and threatened the original proposition.

Figure 1 (page 16) illustrates the current position. In the core environment, circuit switched voice architectures have shifted towards a packet switched ‘data’ infrastructure, creating a core network that exhibits great scope and capacity. By contrast, the access loop has transitioned to an access network but maintained its fundamental voice services infrastructure. There is a bandwidth gap between the two that precludes the realization of adequate revenues from investments in the core.

Thoughts have turned from the function of the network to the form of the network. With present scope providing services to only a small proportion of the global population, but with demand due to that proportion already threatening to overwhelm the architecture, new network forms are being considered. The primary intention: to reshape the ways in which network functions interact and to generate a cascading architecture that provides for more controlled growth and greater user context.

THE INFORMATION GRID

This process has given rise to what is being termed the ‘Information Grid’ or the ‘Great Global Grid’. The Grid model proposes a true network of networks, linked hierarchically from household to global level. Visually, this model resembles a nested distributed mesh — picture several layered sheets of chicken wire, held together at one corner and then lightly twisted to provide a series of nested cones. With this picture, the density of nodes can be seen to decrease from the origin to the edges of each ‘sheet’, reflecting the decline in density of service and traffic flow at each node as the model descends.

Further examination of the model provides evidence that as node density declines, inter-nodal distance increases. This distance increases to the scale point at which the provision of lateral connections between sheets becomes problematic: although these connections would add stability, the amount of material that must be added makes such provision infeasible. Below a certain layer of the model, therefore, node connectivity must be achieved by transiting parent nodes to gain otherwise lateral access to parallel nodes.

In reverse, as node density increases, the distance between nodes declines to such an extent that, above a certain point in the model, the distance is so small as to preclude the requirement for lateral links. Above this point, inter-nodal connectivity can be achieved purely as a result of nodal proximity. The effect of node density within the Grid model thus describes the generation and propagation process for the provision of infrastructure and services within the Grid: the architecture of complexity.2

In direct contrast to the tiered hub and spoke model, the architecture of the Information Grid dictates a requirement for scalable infrastructure.

Figure 2 illustrates the nature of the Information Grid. Each network in the Grid performs a specific role: providing Grid functionality to its direct community and resolving traffic within its own context. Community concentration of this nature allows for localization of resources and content while providing access pathways to elsewhere within the Grid.

THE MESONET

Related to the spectrum of provision generated within the Grid is the idea of a collection of identifiable function-specific ‘soft’ networks, generated dynamically by application instances within a user session. Networks of this type are derived by an application in order to fulfil the requirements of that application. To support this model, the spectrum of provision relative to the Information Grid must include service components, including storage, encryption and authentication, application definable connectivity, or on-command VPN services, that can be used by network applications.

The mesoNET, as Figure 3 (page 18) illustrates, provides for a ‘middle network’. This middle network abstracts the traditional seven-layer OSI model to three layers: physical infrastructure interface, specific application functionality and a middle layer of interrelated network functions. Individual application instances within mesoNET sessions declare a requirement for a dynamic network of functions, generating per function utilization events that may then be billed to the entity instantiating the application instance. The mesoNET model combines network elements to provide a facilitative core in a Grid environment while maintaining traditional functional roles from the OSI perspective.

The mesoNET architecture is not restricted to the network connectivity model. Sub-classing this architecture for additional functionality, from a Grid perspective, illustrates that the same partitioning process delivers benefits to parallel models. Whether such models are storage networks, enterprise computing, or broadcast related, forming the relevant infrastructure from mesoNET principles benefits the implementation and ensures that Grid principles are not compromised.

In a sense, there is nothing particularly new about the idea of a mesoNET: it can be argued that the World Wide Web meets and indeed defines the content of the layers in the mesoNET. Where mesoNET principles differ, however, are in the application of a common structural definition to multiple functional layers of the network. By cascading functional layers according to mesoNET principles, the result begins to exhibit the characteristics of a network arcology: a data space that is fully integrated with its environment.

CENTERNAL SYSTEMS

As the next network gains ground, issues of latency, security, confidentiality, scope, and disproportionate pricing will be directly addressed. As a result, a foundation will be laid that encourages business-critical network deployments. Instead of the current data archipelago, this foundation creates the inverse: a sea of data that places systems, services, and information at the heart of the distributed organization. With this approach, centernal data — data that is centralized externally — becomes available from anywhere, at any time. Flexibility in relation to ideas and responses to market become normal practice and data and communications costs may be managed proactively, if not reduced.

Figure 4 illustrates a basic model for centernal systems. Taking advantage of Grid techniques, dynamic optical connectivity, and mesoNET principled distribution, the centernal model delivers real advantages in terms of functionality, reach, proactive responses and value for money. For the corporate sector, as well as its customers, a centernal m
odel delivers a bigger ‘bang per buck’ than traditional systems implementation.

For those with the ability to take advantage of the model, centernal systems — in conjunction with the Grid and the mesoNET — provide an unprecedented opportunity. According to Vinod Khosla of Kleiner Perkins Caulfield and Byers, a shift in corporate use of technology will generate a new industry by 2005 — an industry worth some US$50 billion (a tripling of present corporate information systems budgets).

One of the ‘dot-bomb’ predictions was that a shortfall in network-related revenues due to that correction would be assuaged by an increase in corporate use. With today’s network, that prediction has yet to pan out but, as critical systems components become distributed within the network, and as dynamic networks are generated to meet application needs, the demand for network capacity will increase. The next network will present benefits to its users, but it also presents a massive opportunity for providers of space, connectivity, capacity, storage, components, and services to that network.

THE NEXT NETWORK

As technology advances, new applications are found that generate a requirement for further advances. The rise of the Internet between 1996 and 2000 is one example of a mass technology application creating demand for components throughout the network: components such as billing systems, commerce engines, servers, network capacity, information services, and colocation space. Supply of such components was duly accelerated, yet the ability of the form of the network to deliver promised functionality to any of the new environments was severely overestimated.

This time, the emphasis is on the form before the function. If we build the right foundations for the pipeline, the opportunities will flow. Watching the emergence of the Information Grid, mesoNET functionality, and corporate ‘centernalisation’ leads to a single conclusion: the next network will provide a distributed, critical mesh linking intelligent, functional nodes.

Figure 5 presents the main elements in the next network, showing that new models abound where ideas overlap and intersect. Despite market sentiment and the downward trend for technology and communications, there is a groundswell of anticipation that the next network is coming — a network that is adaptive, visual, polymorphic, and of which the adoption curve will be nearly vertical. Elements of this network are already in place, but what remains will define the outcome of a 20+-year journey through information systems and technology.

Completing this journey requires work. Continued education, investment, adaptation, and belief are all necessary to effect a transition to the next network. Issues of models, mindsets, and methods will have to be addressed. This barrier will be difficult to surmount but, with application and effort, the rewards will be plentiful.

David Prior is the founding director of Prior Intelligence Ltd., a UK company specializing in the provision of strategic intelligence with regard to the components of the next network. This article is drawn from a white paper entitled ‘Beyond Colocation: Cornerstones of the Network Arcology’, available on request from enquiry@priorintelligence.com.

Footnotes

1 Khosla, Forbes, etc. – Historical analysis of business, finance, and technology cycles

2 Term used by Barabasi, Albert-Laszlo: University of Notre Dame, Indiana

BY DAVID PRIOR


Print this page

Related