Modes of Computing veryard projects > service orientation > computing

Autonomic Computing veryard projects > service orientation > computing > autonomic

IBM’s work on autonomic computing has come out of the eLiza research project, and is starting to be trickled into the IBM product range via Tivoli. Other vendors with products in this space include Sychron Software and Opalis. Microsoft's more recent Dynamic Systems Initiative is also addressing some similar requirements.

Complex adaptive systems

Most of the autonomic functions of the human body are outside conscious control. Your body protects itself against minor infections without consulting the brain. However, there are various levels of system management at which conscious intervention is possible – for example, allowing you to respond to the early signs of a cold by taking vitamin pills or visiting the doctor. An autonomic computer network may generate signals (sometimes known as algedonic signals) which indicate the need for human intervention.

Autonomic computing addresses security

Of course, there are already lots of products with fault-tolerant and fail-safe features – and not just in computing hardware. Safety features based on these ideas can also be found in motor cars – from anti-lock brakes to run-flat tyres.

In simple systems management, things either work or they don’t. Security checks are binary – things either pass or fail. In contrast to this, autonomic systems allow for partial working, provisional checks, with ongoing monitoring and adjustment. This means they can respond positively to a much greater range of environmental conditions.

Management of complexity

The notion of self-managing systems is both an ambitious one, and one that can be seen as a natural progression from notions of resilience and fault-tolerance that have been common in both computer hardware and telecommunications networks for many years.

Autonomic computing involves making the individual components and services more manageable – which has implications for their design and implementation – as well as providing a separate management layer. For example, IBM envisages a self-managing wrapper around each component – and these wrappers may communicate peer-to-peer, as well as with a central or decentralized management function. The manageability interfaces will (of course) be rendered as web services.

Configuration and optimization

Other IBM products have self-optimization and self-healing functionality. For example, the latest version of Websphere includes distributed workload, with failover and hot deployment.

HP offers some automatic resource allocation functionality in its UDC (Utility Data Centre) product range – so that resources can be allocated dynamically as required, with appropriate billing function. A hardware vendor such as HP can ship boxes with dormant processing power, which can be activated (and automatically billed) only when the extra processing power is actually used.

More advanced functionality is currently in development. For example, HP is working on a framework for the development of configuration-driven systems called "SmartFrog" (Smart Framework for Object Groups), which will include a configuration description language to describe the precise, desired configuration of applications composed of sets of components running across a distributed system, and a distributed deployment system to realize application descriptions, and to monitor and manage the resulting applications through their lifecycles.
 

vendor material	IBM webpage on autonomic computing Opalis Sychron Software
analytic material	CBDi material, January 2003 Veryard Projects material: IBM, Model-Based Management
other material	Complex Adaptive Systems

veryard projects > service orientation > computing

Page updated November 22nd, 2003 Copyright © 2003 Veryard Projects Ltd  http://www.veryard.com/so/computing.htm