Enterprise networks are increasingly complex. In fact, the largest of them may support tens of thousands of users, thousands of sites and hundreds of applications. Traffic patterns are constantly changed by new architectures (e.g. data center consolidation, desktop virtualization), new applications (e.g. peer-to-peer, voice, telepresence) and even time zones. Non-critical traffic increases exponentially, threatening business applications’ efficiency and putting network architectures and budgets under pressure.
On the other hand, traditional traffic management principles and tools that network teams use (so-called policy-based management) have outlived their usefulness. With them, partial information is collected on local devices (e.g. routers, probes) and is then manually processed by overwhelmed engineers that must deduce approximate local policies to be entered in such devices.
This cumbersome process is not able to match the challenges of speed and complexity that are facing agile and global enterprises. It is time for a new, high-level traffic management concept that delegates to processors what they do best: exchange information and compute fast and accurately, while preserving human intelligence to set-up the mission (objectives) and control its achievement (reporting, service level agreements).
Autonomic networking is a fundamental step to manage the ever growing complexity of enterprises’ network and application environments.
When is a system is autonomic? If you must tell the system everything it has to do, then it is not autonomic (it’s probably policy-based). On the other hand, if you tell the system what mission to accomplish and then let it manage the on-field situations, then it is an autonomic system.
Autonomic networking defines a new generation of intelligent systems that make decisions according to the context in which they evolve. In the WAN optimization area, an autonomic networking system offers the following benefits:
- Adapts its behavior dynamically and in real-time based on a continuous assessment of available network resources, immediate user demand and high-level objectives;
- Is based on distributed computing principles in order to be fast enough, extremely resilient and able to scale up to the largest networks;
- Leverages collaboration between decentralized agents to take into account correlations between local events and reach global efficiency.
An autonomic networking system delivers the following strategic advantages to an enterprise:
- Managed from a tiny number of parameters, it is extremely easy to deploy and to configure;
- Its high level of automation keeps the maintenance and change management workload several order of magnitude below old-fashioned policy-based systems;
- Always tuned, it is able to delivering continuously the best performance even in the most complex and dynamic network environment;
- Nearly independent from the actual network size as well as from applications and user traffic, it is extremely scalable;
- Self-healing, it does not exhibit single points of failure.
With policy-based systems outmatched by the complexity of today’s networks, autonomic networking systems are needed by businesses to meet their organization’s advanced requirements and remain competitive.
Thierry Grenot, founder and Chief Technology Officer at Ipanema Technologies, writes the WAN Optimization column for TMCnet. To read more of Thierry�s articles, please visit his columnist page.
Edited by Erik Linask