A fundamental tenet of autonomic computing is to increase the intelligence of individual computer components so that they become "self- managing," i.e., actively monitoring their state and taking corrective actions in accordance with overall system-management objectives. The autonomic nervous system of the human body controls bodily functions such as heart rate, breathing and blood pressure without any conscious attention on our part. The parallel notion when applied to autonomic computing is to have systems that manage themselves without active human intervention. The ultimate goal is to create self-managing computer systems.

The motivation for autonomics is one of rapid growth and complexities in computer systems. Annually, the number of connected computing devices is expected to grow at 38% [1]. To manage this complexity, the human labor cost is exceeding equipment costs by a ratio of up to 18:1 [1]. This results in complexity in the computer networks making them hard to control manually by human operators. The economics of this highlights the necessity for autonomics in computer systems.

The six papers in this issue of Intel® Technology Journal (Volume 10, Issue 4) review in depth the research work at Intel Corporation on autonomic computing, an important direction for future computing.

Platform support of autonomic computing: an evolution of manageability architecture

The first paper explains the Intel^® technologies that provide platform support for autonomics. Specifically, these are computer platforms with sufficient support and on-board intelligence to enable autonomic capabilities, such as self-healing and self-protecting, as well as discovery and asset tracking. To achieve this, dedicated platform resources and firmware with well-defined standard interfaces implement a set of management and autonomic capabilities. This paper also explains the Intel^® Active Management Technology (Intel AMT), which is the first Intel product that supports autonomic computing.

Service orchestration of Intel-based platforms under a service-oriented infrastructure

The second paper describes research on Service-Oriented Infrastructure (SOI) that enables higher-level service orientation and autonomic computing. We demonstrate how the concept of platform as a service (PaaS) may be applied to real-world Information Technology (IT) operations. The results show that SOI is viable and PaaS is achievable.

Standards for autonomic computing

In the third paper, we highlight important standards that enable components from heterogeneous sources to interact with each other. This interaction is fundamental to enabling intelligent decision making at the lowest possible level in the systems management hierarchy. We describe the standards required for external interfaces for autonomic elements such as Web Services (WS) Management and WS Distributed Management. We also explain the general design philosophy of these two approaches.

Towards autonomic enterprise security: self-defending platforms, distributed detection, and adaptive feedback

In the fourth paper, we describe three building blocks that help realize the ultimate goal of "autonomic operation" of the enterprise. First, we describe the notion of self defending platforms which enable an end-host to detect program-level anomalies and unauthorized modifications. Next, we describe the concept of distributed detection and inference, where end-hosts collaborate regarding the state of the entire network. Finally, we discuss an adaptive policy management architecture that supplements the self-defense and distribution detection capabilities.

Machine learning for adaptive power management

In the fifth paper, we describe how a machine-learning methodology could be applied to adaptive power management. We propose a system that learns when to turn off components based on user patterns. We describe the challenges of building such a system and explore a range of solutions.

A self-managing framework for health monitoring

In the sixth paper, we propose a self-managing framework for health monitoring using body-wearable bio-devices that can reduce the doctor’s intervention for patient management. We illustrate three usage models where this self-managing framework can be applied: remotely monitoring patients at home; monitoring fetal well-being in a maternity ward; and monitoring critical patients in an Intensive Care Unit (ICU).

Today, as networked devices grow rapidly, the complexities of managing them necessitates that we work both in industry and in academia to explore how to build modern, networked computing systems that can self-manage. Research in this complex area of computer science can help foster a whole new type of computer systems and capabilities for the next decade.

[1] Wikipedia on autonomic computing http://en.wikipedia.org/wiki/Autonomic_computing