Cognitive Networks: Towards Self-Aware Networks

4.1. Introduction: Concepts and Challenges

Human reliance on advanced digital communications and distributed applications has reached a phenomenal stage, where everyday life and business processes hinge on the proper and efficient operation of computer networks and applications. It is apparent that networks and software systems are becoming larger in scales, more complex in design, more pervasive in distribution, and the runtime environment more dynamic – they have grown beyond the limit of manual administration. Yet at the same time, there is rising pressure in making networks and systems more manageable, their operations more efficient, and their deployment and maintenance more cost effective. In 2002, IBM stated that the IT industry is facing a crisis and network and system management is becoming the grand research challenge ^[]. The immensity and immediacy of this problem is perhaps best illustrated with today's stark IT business reality ^[]:

• 33% of the total cost of ownership is spent on network and software recovery and 50% on failure protection.

• 40% service outages are caused by human error.

• 80% IT expenditure is on operations and maintenance.

As a result, researchers have been actively seeking solutions to this management crisis. Although advancements in hardware and software technologies (e.g., decreases in computing and network capacity cost, emergence of distributed software, web service technology) will ameliorate the problem to some extent, the ultimate solution lies ...