Pipeline Publishing, Volume 4, Issue 3
This Month's Issue:
download article in pdf format
last page next page

Autonomic Networks - Autonomic Communication

back to cover
article page | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |

When grids become the platform of choice for deploying services, and open networks can access grid based services, the re-identification of computer and network will be complete. We see the start of this today in two landmark companies and it is no coincidence they are both models of success for the internet economy. Akamai is a grid of tens of thousands of distributed servers that spool content to edge users – Akamai provides parallel downloads at points that are ‘network close’ to the end user. Google provides an ever growing number of consumer services hosted in parallel on what many believe to be the world’s largest grid – on the way to hundreds of thousands of servers. These companies are unified service, computer, and network platforms. Already, the financial powerhouses, based on the success of their use of grids as massive computation engines, are studying moving all their future services to grids. Where goes the financial service industry, soon goes the rest of industry. Grids, Pervasive Computing, and Ubiquitous Networks are likely to be common in one decade. This places pressure on turning the study of autonomic networks into the applied deployment of autonomic network infrastructures as quickly as possible.

Since we are re-inventing not just networks, but services and organizations, a broader name for the activity is needed - one both comprehensive and immediately clear.

Autonomic Communication (AC) - a new communication paradigm to assist the evolution of communication networks towards functional adaptability, extensibility and resilience to a wide range of possible faults and attacks. Special emphasis is given on the grounding principles to achieve purposeful behavior on top of self-organization (including self- management, self- healing, self- awareness, etc.) … in its interaction with numerous often-dynamic network groups and communities. The goals are to understand how autonomic behaviors are learned, influenced or changed, and how, in turn, these affect other elements, groups and the network. “

This term, Autonomic Communication (AC), seems to us to best summarize the field of activity. Hereafter we refer to the paradigm shift and research as Autonomic Communication (AC) and the product that will be deployed as the Autonomic Network (AN).

Architecture of Autonomic Networks

Not everything is new in transitioning to Autonomic Networks. The current thought on network structure remains. For example, autonomic communications holds the layers as:

The major components have been identified, groups of talented people are focused on task, and there are new revenue models waiting in the wings.

  • User
  • Applications
  • Services/ Components
  • Middleware
  • Network & Computers
  • Device
  • Agent(soft) or Sensor (hard)

Notice however that this layering includes network, computers & software, transmissions, and information. So while these are familiar, what was once several distinct groups are now combined.

From the requirement of supporting Ubiquitous Networks, a strong domain model must be incorporated in Autonomic Communication architectures. Domains replace layers as the primary organizing unit. Domains will provide for semi-permeable boundaries, much like the edge & gateway routers do today in IP networks. But domains will need to automatically communicate large amounts of information about what they will and will not accept; what they are capable of, and what current capacity and QoS is available. Intra-network control- communications- paths will need strong security structures, gauged to how open or how restricted is the information context of the services they accept. As we described before, here is a fractal-like repeating pattern. Domains will contain domains will contain domains - Each with similar computational control systems but with different locations, contexts, and communities. Pan domain structures will link and provide shortcuts across these numerous networks. Indeed, something like the old I-PNNI model may return.

A distributed computing substrate (including tools and platforms) will be an embedded part of the network. While routers and switches will still pass data, they will outsource all but the simplest and quickest of decisions to the embedded distributed computing substrate; only things like lookups will stay. Indeed, the concept of a management interface fundamentally changes from a ‘passive responder to queries’ or a ‘broadcaster of alarms’ into an ‘active interface’ which links to control systems and network/service models. In effect, every device will have a service interface with published methods. In some architectural systems, the device-embedded service will actively register itself with service modelers and discovery components – principal of which is the agent-service providing self-restoration. This probably means that every device will incorporate Java or .NET and provide service and management interfaces accessible by soft services.

article page | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
last page back to top of page next page

© 2006, All information contained herein is the sole property of Pipeline Publishing, LLC. Pipeline Publishing LLC reserves all rights and privileges regarding
the use of this information. Any unauthorized use, such as copying, modifying, or reprinting, will be prosecuted under the fullest extent under the governing law.