Slashdot Mirror
Initiative for Autonomic Computing Gains Strength
museumpeace writes "Tired of fixing your computer? What if your system broke down two billion
miles from the nearest spare part or human? NASA has just held a
colloquium where Ulster University computer science researcher Roy Sterritt was invited to present his ideas on Autonomic Computing. In the last few years,the leading system vendors have realized 'There is no less than a crisis today in three areas: cost, availability and user experience.' There has been a fair amount of academic research since customers like NASA see in it the potential to make remotely operated complex systems sustainable. It all makes for some very cool systems design work and there are lots of further research opportunities. Just don't forget what it may do to your job."
Yah the leading system vendors have realized there's a crisis. How else are they going to sell more systems if the ones in place now aren't dangerously unstable? They could probably explode at any minute, are toxic, and will probably delete all my data at any second.
I better go buy a new computer.
Automated nanobots
Now we need only worry about the whole thing going berserk, killing the crewmembers, and attempting to destroy the Earth.
JMD
When all else fails, feel free to panic.
If self-fixing computers become the norm, that means half the phone calls I get from friends will stop.
Hmmm....bug or feature?
It won't matter until it can fix user errors anyway.
This is another way of starting a sig with this and ending it with that.
It's just about impossible that a tecnic that makes robotic spacecraft all that much more self sufficient will be confined to just robotic space travel for long. If NASA is successful, we will see widespread robotization here on Earth as a consequence.
30 years from now, this will be characterized as a 'mere spin off', and instead of bitching about Moonrocks, ignorant people will be saying "We spent billions to send robot probes to Pluto, and all we got was a bunch of contaminated Helium."
Who is John Cabal?
grep -c icrosoft *
In the ZDNet article on Google's inner workings that was posted earlier on /., Urs Hölzle mentions that in the larger Google clusters, 2 machines per day will fail. They compensate for this with triple redundancy, good software for failover control, and a staff of 800(!) computer scientists. Needless to say, not everyone could manage this... there's definitely an enterprise niche for system autonomy. This also brings IBM's eFuse technology for self-repairing chips to mind.
perl -e 'foreach(values %SIG){$_="IGNORE";}while(){}'
For years SAN's from EMC, fault tolerant serves from Stratus, etc. have all had the ability to phone home when they detect a failure is imminent or has occured. Usually the customer doesn't realize there's even a problem until a service tech shows up with replacement parts.
Of course getting this down to the level of home users is still a long way away...
I for one welcome our new miniature overlords.
It was a joke! When you give me that look it was a joke.
There's a race, Manufactuer's building smarter computers and AOL signing up dumber users.
So far AOL is winning
code name for this project 'SkyNet' by any chance?
Fix moved/broken links:
0 15 2-8C15-1CDA-B4A8809EC588EEDF&pageNumber=3&catI D=4 cites:o ns
/public/publications was not found on this server."
http://www.sciam.com/article.cfm?articleID=000B
http://swig.stanford.edu/public/publicati
"Not Found
The requested URL
People have been trying to make systems easier to manage for years. Unfortunately, it's not enough to have the desire to make systems self-managing, you also need good ideas for how to do it, and those are still lacking as much as they always have been.
Give the guy credit, though, for seeing a good opportunity. Industry will believe in this silver bullet like they have done in the ones before.
Unfortunately, the real research will still take decades to complete, and then this area will have a bad name just like most of the other overhyped technologies before it.
I have been talking about this for years...
If the autonomous systems NASA and the ESA have put into the void are any indication, I don't think we have much to worry about - the costs will be prohibitive for all save the largest organizations, and true autonomy (in the form of robotics) will have a whole range of other problems (imagine your main file server getting up and walking out of the data center because it mistakenly assumed there was a fire...)
The key, in the interrum is make yourself indispensible. If you have the mindset that you are a code grinder/monkey and that is all you want to be, then your days are numbered. Your goal should instead be becoming the guy who can put together a complete solution (data, application, hardware, network) in short order that works, scales well, and is extensible by your users. You need to be a jack-of-all-trades. That is how to survive and gain esteem in the eyes of your clients and peers, as I see it.
Lodragan Draoidh
The more you explain it, the more I don't understand it. - Mark Twain
I wouldn't consider this to be new...rather it's the idea of this that is starting to propigate.
CISCO's new 92 terabit/sec router already has some of these features. The OS they used to build the system supports many of these features (high availability, self healing, etc).
http://www.qnx.com/markets/networking_telecom/cisc o/
http://www.cisco.com/en/US/products/ps5763/index.h tml
It's a self healing system. It uses the services and functionality of the OS to accomplish it.
QNX's networking system is really neat because it allows processes to be independent of where they actually run on a network. And the network can be anything (i.e. a backplane, Ethernet, whatever). So it lends itself to solving such a problem.
The IBM links says, under "The Solution":
In conventional system design, the Rs of reliable systems are: (1) Robust, (2) Repair, and (3) Redundant.
Biological systems use all three methods to varying degrees but the problem is that biological systems do not survive as individuals, they survive as a species by tolerating a high degree of failure and using a fourth R: Replication.
For computer systems, this biological systems approach would mean replacing every component of the system on a regular basis the way all the cells in the human body are completely replaced every seven years. Periodically, you would throw out the entire system and replace it with two or three new ones that have undergone a period of testing and development.
The replication approach, which is key to the survival of biological systems, runs counter to most business thinking, which is to replace multiple systems with fewer, more powerful systems. This limits reliability to the first three Rs.
There is much that can be done to increase reliability with these 3 Rs but if biological systems are any indication (as well as some theoretical limits), they are inadequate.
The problem of reliability could ultimately be a flaw in the way business works rather than a technical problem.
As we modelled the eye to build cameras, the brain to build computers, the ear to build speakers, we're modeling our autonomic nervous system to build the next evolutionary step in computing. Networks that independently and reflexively self -regulate, configure, repair, optimize, and protect in the same sense as an immune system or an automatic pilot.
This would allow the network to automatically manage server load balancing, process allocation, monitor the power supply, automatic update software and fend off threats without having to consult the administrator.
For example, if an application starts performing badly, it automatically receives increased resources. If software or hardware fails, it doesn't even ripple the end users coffee. An autonomous computing system would roll out new patches, monitor and adjust the resources singular end users need, set up servers... all the mundane stuff.
The complexity of integrating and managing the latest hardware and software into existing systems is destroying the advantages of economies of scale. Autonomic computing is one way of insulating the IT administrator from the mundane complexities and freeing them to do other more interesting things like understanding the needs of the business more, or modelling and automating existing business processes.
On a larger scale, it spells an evolutionary move towards a decentralized global self-configuring, self-healing, self-optimizing, and self-protecting nervous system. Since Autonomic Computing can look for patterns in data and extrapolate to predict future events, deployed on a global scale, the spin-offs would be very interesting.
~~~~~~~
Thoughts on the Emergence of Computing Intelligence
What if your system broke down two billion miles from the nearest spare part or human?
I think they'll do a one-day deliver on this for a small surcharge.
Dedicated Cthulhu Cultist since 4523 BC.