|
SERVICES
System, heal thyself
Vital Infrastructure avoids downtime
by Scott Steinberg
As any IT professional will tell you, system failures are regrettable. Are they unavoidable? Well ... that's another story.
With even a single hour of downtime capable of crippling an organization's functionality, it's clear that a data warehouse is an investment that must be protected. Seeking a proactive solution to downtime concerns, today's corporate leaders are increasingly turning toward the concept of system self-healing.
Like many procedural developments, the self-healing concept is founded upon a technical cornerstone; in this case it's Teradata's Vital Infrastructure package. Teradata's Customer Services team works with companies to create a predictive environment that requires less downtime.
By looking at all parts of a system as opposed to single components, the system can detect a unique enterprise signature. Rather than simply collecting and categorizing event and state data separately, self-healing centralizes serviceability information and interprets correlations. Afterward, corrective actions are proposed, tested and taken when a predefined knowledge base diagnoses a possible problem area. In doing so, the software checks current system signatures against a collection or database of known patterns and trends—safeguarding against potential flaws before they occur.
"The principle hinges on creating a reliable and robust predictive environment that becomes increasingly autonomous," says Douglas Irion, CTO of Teradata Customer Services. "Putting the proper architecture in place for adaptive learning is critical ... as a preventative measure, it makes all the difference." Given that downtime avoidance is the objective, a service-oriented architecture that lies at the model's heart becomes the key factor in crisis resolution. Why? Having a framework in place that lets governing rules be modified with haste—and on the fly—is crucial, as firsthand experience shows.
Although it operates in a methodical manner, self-healing is largely an iterative approach. After being notified concerning probable errors, analysts and engineers provide real-world feedback on the scenario. Observations are then incorporated into system analytics. The results culminate in an intelligent fault-detection engine that matures over time.
Simultaneously, a working roadmap is generated and placed at field representatives' fingertips in the form of browser-driven applications. The bottom line? Data previously hidden from event handlers is now freely available, giving the end user a shot at addressing problems internally, thus saving time and money.
"Our destination is a self-aware Teradata system—one that can recognize random events that may lead to problems and automatically initiate a fix," asserts Craig James, program manager, Teradata Vital Infrastructure.
The promise of such practices is evidence of an ongoing push to create self-awareness within systems. In fact, a 2002 Gartner focus report, "Software Support Companies Implement Support Automation," predicts that 75% of software support incidents will be resolved by support automation technology before 2005.
While the concept remains in its early stages, associated payoffs are already manifesting themselves in the areas of incident avoidance, deflection and redistribution. Across the board, days-to-resolution and time-per-incident figures have been slashed. Uptime is also steadily increasing. Perhaps most rewardingly, what were once viewed as difficult problems are often now no more than moderate or minor setbacks.
For business owners who are bent on staying at the forefront of their respective industries, the choice between embracing the technology or watching as costs skyrocket and productivity exponentially plummets is obvious.
"Our goal is to improve our customers' productivity through continued investment into the manageability and predictability of the Teradata system," says Antonio Nevarez, director, Teradata Services Product Management. "Through delivery of self-sufficient systems, it's like always having an expert on-hand."
Scott Steinberg's articles have appeared in Wired, Popular Science and Rolling Stone.
© Teradata Magazine-June 2004
 RELATED ARTICLES: |
|
|
|
|
|
Printable version
Send to a colleague
