Self-regulating, or autonomic, computing is an idea that vendors say is ready to leave the research lab and set up shop in mainstream software.
At a recent panel discussion on autonomic computing, hosted by IBM at the companys Innovation Center for Business Partners, in San Mateo, Calif., experts said that self-regulating systems will gradually enter the workplace—and that some products already include autonomic features.
As computers have gotten faster and more powerful, enabling more data to be processed, it has become harder to understand what they are doing. In turn, systems have become commensurately more difficult to manage and more expensive to maintain. According to most industry analysts, from 60 percent to 80 percent of current IT budgets are spent on system maintenance, leaving organizations with few resources for innovation.
Autonomic systems—networks that can scrutinize, administer and repair themselves—have the potential to alleviate some of that IT staff/budget burden. Self-regulating systems, vendors said, could also lower computing costs by allowing IT staff to have a macro focus on their technology at hand, rather than having to micro-focus on bits and bytes.
No vendor has pushed this strategy harder than IBM. During an October 2001 Agenda conference, Paul Horn, senior vice president of IBM Research, challenged IBM as a corporation—and the entire industry—to make computing function more like the human bodys autonomic nervous system.
Autonomic computings status as a catchphrase may be new, but IBM has long pursued self-regulating features within products such as its DB2 Universal Database.
eWEEK Labs first saw glimpses of this strategy as far back as DB2 Universal Database 6 for Unix, Linux and Windows. This release featured the Index Advisor tool, which automatically determined the best set of indexes for a given workload and space constraints.
Since then, we have seen IBM extend the technologies of DB2s Index Advisor with the introduction of Design Advisor, which identifies all the objects needed to improve the performance of a workload, in DB2 8.2.
“What weve tried to do is turn this big ship called computer science from what weve spent the last 50 years doing—developing faster and faster machines,” said Guy Lohman, manager of advanced optimization for the Advanced Database Solutions Department at IBMs Almaden Research Center, in San Jose, Calif., during the panel discussion. “Its time to change what were working on. What we will work on for the next 50 years is making systems work.”
The research center continues to study autonomic computing. Lohman, for example, is working on problem determination and said the IBM labs have a pipeline of projects in development as features for DB2.
“[One day,] there will be a hub of communicating, intelligent autonomic managers constantly monitoring, analyzing and taking actions in response to the goals you set for your IT system,” said Ric Telford, director for autonomic computing at IBM, during the panel discussion.
IBM isnt the only vendor focused on self-regulating computing, of course. Microsoft Corp. has introduced repair features in recent versions of Office, allowing corrupted or accidentally deleted program files to be automatically fixed and reinstalled by the software.
Researchers at the University of California, Berkeley, are working on a software project that is similar to autonomic computing called ROC (recovery-oriented computing), while researchers at Hewlett-Packard Labs are working on their own version—which they call planetary computing.
Although most systems have evolved beyond what a small team of humans can operate, the goal is not to take humans completely out of the loop. Administrators should, and will, continue to approve any action taken by systems and to define system goals.
Said Lohman, “All this may be low-hanging fruit, but weve got to start somewhere. Are we ever going to get to the point where we dont have any humans in the loop? No.”
Self-regulating systems may never maintain themselves to the extent that living things can, but the goal is to create a hub of communicating, intelligent autonomic managers constantly monitoring, analyzing and taking action in response to parameters set by IT managers. Vendors might be taking baby steps right now, but autonomic computing is clearly an idea thats quickly coming of age.
Senior Writer Anne Chen can be reached at [email protected].