Pipeline Publishing, Volume 4, Issue 4

	This Month's Issue:
	Maintaining Network Health

Network, Heal Thyself!
Technological Advances Put Building Blocks in Place to Create Self-Healing, Reactive Network Architecture

article page | 1 | 2 | 3 |

The policy server accomplishes this by first deconstructing the end user request into the specific service components that require action or changes to be made. This determination is made possible by a comprehensive repository of networked resources that includes all the deployed devices, available network services, and resources maintained by the policy server. From this repository, the policy server is able to determine all the deployed resources that require changes to satisfy the end-user request.

Because the network must be able to react to outside occurrences, the repository must be able to define and manage resources external to the network such as application servers, desktops, laptops, and users. These external resources are modeled in the repository in the same way as the network devices.

Once the resources that need to be re-configured are determined, the policy server sends directives down to agents running on the actual devices. These directives are constructed using a markup language, which is interpreted by the agent into the device-level command language necessary to initiate the services through the SOA layer on the device.

The combination of this new generation of policy servers – which can define, control and implement desired behaviors in underlying network devices – with the new open architectures and exposed SOA layers on these devices allow for dynamic and instantaneous changes to the logical network environment and behavior.

The final building block for our self-healing network solution consists of a sophisticated rules engine that defines events in the network, or external to the network, and the appropriate actions to take in such a form that the policy server can act upon them.

This rules engine must be capable of sophisticated logic patterns so that complex conditional decisions can be made. The ability to create complex rules, such as “if A occurs, then do B, but if A and C occur, then do D,” are necessary to effectively define a network capable of responding to real world events. In the real world, there are very few events and actions that consist only of a singular event-to-action relationship. Most of the scenarios network engineers deal with on a daily basis have multiple contributing events and multiple actions.

As with the policy server, the rules engine in our solution has to be capable of modeling events and actions outside of the network. As the policy server receives notice from an agent residing on a network device or non-

As every network engineer understands, any change introduced into a system can drastically impact the behavior of that system in any number of ways – good or bad.

network device (server, desktop, laptop, or even a handheld device) that a defined event has occurred, the policy server queries the rules engine for the appropriate response(s) to the event and begins to adjust the underlying network devices accordingly.

Obviously the complexity of the dynamic configuration or self-healing solution resides in the proper definition of events and actions. The mechanics of dynamically implementing change into the networked environment are relatively minor compared to correctly defining and controlling the change behaviors being implemented. More importantly, as every network engineer understands, any change introduced into a system can drastically impact the behavior of that system in any number of ways – good or bad.

To minimize the risk of building a house of cards that may collapse upon itself as a new change is being introduced based on an event, the policy server must maintain a persistent linkage at all times between deployed configurations and the policies that implemented those configurations within its repository. This critical functionality allows the network engineer to model potential event scenarios and see how the underlying logical network will behave as those incremental changes are being introduced, without disrupting the actual production networks.

Using these building blocks in conjunction with many other existing network technologies, we are finally making fundamental advances in network automation and self-healing. Companies working with these technologies are currently deploying dynamic network solutions such as the emergency response network described earlier. And, with further refinements from leading hardware and software companies, network engineers will soon have the ability to control and manage any network infrastructure using defined policy, and have that network be capable of recognizing events and implement appropriate actions automatically across the network fabric as needed. A network that can truly heal itself.

article page | 1 | 2 | 3 |

© 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.