The Hindu Business Line : Computer, heal thyself

During Christmas time, I bought a printer-cum-fax machine and with a lot of effort, configured it to work. My dad, a retired professor of IIT who got his doctoral degree using computers and punched cards, asked me "Do we have to spend so much time to set it up? Can't it be like a refrigerator that you just plug in and it starts working?" I tried to explain to him in vain.

Why not it be made simple?

Welcome to autonomic computing. Like a human body that knows when to breathe, computers that can Configure, Heal (repair), Optimise and Protect themselves are autonomic computers. They have self-CHOP characteristics.

IBM first announced this in 2001 and has done pioneering work in developing and incorporating this into its products. Since then, Microsoft (trustworthy computing), Sun (N1 initiative) and Hewlett-Packard (adaptive computing) are doing research in this area.

What capabilities do autonomic computers have?

Self-configuring: They (hardware and software systems) have the ability to define themselves "on-the-fly" i.e., new features, components, computers, can be added dynamically with little or no human intervention or interruption.

Self-healing: They can detect failed components, machines, repair them and reintroduce them without any disruption.

Self-optimising: They use all the resources (computing power, memory, storage, databases, etc.) optimally.

Self-protecting: The systems can defend themselves against virus attacks and unauthorised access, detect intrusions and report them, and do automatic backup and recovery.

These computers function in a way similar to a human body's involuntary nervous system that adapts the body to changes in the environment and protects it from viruses.

IBM defines the following eight characteristics of an autonomous computer.

Is aware of itself.

Anticipates users' actions.

`Talks' to unfamiliar systems.

Uses open standards for communication.

Tries to continually enhance performance.

Heals when `hurt'.

Protects itself from `enemies'.

Adapts to changes.

IBM classifies the IT environment in a business into five levels of maturity.

Basic: IT professionals individually monitor each system, repair and replace each as needed. The focus is on the time needed to complete major tasks and solve major problems. This involves an informal, reactive and manual process.

Managed: A network management system such as Openview (HP), Tivoli (IBM) or Unicenter (CA) collects and consolidates information about different systems and displays on a few consoles. The focus is on the availability of resources, time to complete formally tracked requests. This involves using a few industry best practices, and a manual review of IT performance.

Predictive: Here systems are monitored and information from sensors is analysed. But the human user decides the course of action. The focus is on availability, performance, and Return on Investment (ROI). The process is proactive and has short approval cycles.

Adaptive: These systems monitor and take actions by themselves. The focus is on end-to-end response times and ability to adapt to varying workloads. The process is automation of many best practices.

Autonomic: The financial implications of all activities are analysed and the IT systems are governed by business policies and priorities. The focus is on optimised performance, competitiveness of service level agreements (SLAs) and ability to deploy new solutions. The process is total automation of all best practices in service and resource management.

Working through feedback control

How do autonomic systems work? They do not use artificial intelligence or neural networks but a feedback control. They use `sensors' and peruse through the system logs to identify a problem. To facilitate this, all the system logs should be written in a common format. IBM has developed the Common Base Events (CBE) Format, an XML-based vocabulary. As part of the autonomic computing Toolkit, IBM provides a Generic Log Adapter to convert logs of mainframes to CBE Format. There is also a Log and Trace Analyzer that provides a graphical interface to view events from the logs of different applications.

The user can define business rules/scenarios and tasks that the computers need to perform when those scenarios occur. For this, IBM has developed an orchestration product. Additionally, the systems must be optimised for best performance. An Enterprise Workload Manager (EWLM) is used to manage heterogeneous systems (different operating systems). Again, a standard format called the Application Response Measurement (ARM) is used to gather performance information from heterogeneous systems because each system will present the information in a different format.

All these calculations involve complex logic. One can use JavaBeans components that have "artificial intelligence-like capabilities" to analyse this logic. IBM is also developing a new tool called, `Agent Building and Learning Environment (ABLE)', that has different types of "AbleBeans" (similar to JavaBeans), which can be combined into "AbleAgents" using rules. In Microsoft's parlance, trustworthy computing has four pillars — reliability, security, privacy and business integrity. Microsoft has integrated anti-spam and anti-phishing technology into MSN, MSN hotmail, Microsoft office 2003 and Exchange Server 2003.

Windows XP and Office XP automatically capture details about software or service failures and give the user the option to report them. (Many of you using Windows XP might have seen a dialog box with a message, "Do you want to report this error?" on your home PCs when something has gone wrong). Microsoft analyses the error data statistically, identifies the fault-patterns, and the solution is posted on the Microsoft Web site.

Sun has built in self-healing features into its Solaris 10 Operating System. They can act prophylactically to "isolate and disable a faulty component and continue providing service even before human administrators know that there is a problem. These components can be dynamically loaded, unloaded and upgraded on-the-fly while the system is running without requiring downtime and without losing any of the active diagnosis state."

Uses of autonomic computing

What kind of tasks can be performed by autonomic computing? It can handle routine maintenance tasks such as installation of software, provisioning of new drives, configuration changes and patch installations. These are only a few, but the list can continually grow. Autonomic computers can't become sentient but close enough to protect themselves, and if `fallen sick', diagnose and take `medication'. This technology is in the nascent stage. Gartner estimates that it will take another five to seven years for it to mature and be used widely.

(The author is a consultant with Satyam Computer Services Ltd.)

More Stories on : IT Training | Hardware

Article E-Mail :: Comment :: Syndication :: Printer Friendly Page

Stories in this Section
Tete-a-tete with experts on the telecom verdict

Scoring in the test

Your coffee - and your bill - will be served in a jiffy

For the book lover... Read away

Measuring your ego in three easy steps

Computer, heal thyself

USB 2.0 drive popup message

`Elixir' of business

Unfolding action

By word of mouth...

Quiz

Go beyond the hub-and-spoke model

Cartoon

Convincing argument

Track your vehicle