Autonomic computing at the Friday paper seminar

Autonomic computing is one of those things that involves a certain amount of hand-waving but that corresponds to a really simple idea: tell a computing system what’s expected of it, instead of a step-by-step account of what we want it to do. We can tell a network we expect 99% uptime, or a database system to improve its performance by 5%, or a webserver to try and serve ten thousand simultaneous connections. Most systems will look back at you slack-jawed, but, well, that’s why the paper we’re dealing with today is called Research challenges of autonomic computing, which is written by Jeffrey O. Kephart, an IBM researcher which has produced lots of fine papers on the topic.
What are, then, the challenges and how does all that relates to our own research? The idea is to achieve what are collectively called self-star properties: self configuration, self-management, self-protection and self-healing. That is, let the system itself handle as a body, with each component being like an organ that performs its own function but is also aware that must work for the collective good and obtain an emergent property like body temperature or oxygen supply. That double function leads to architectural challenges, but also to algorithmic problems like how to be aware of what the system at large is doing and how it is behaving, how to learn new responses to changes in environment, and, eventually, how to optimize the system configuration and behaviour to attain desired targets set by the user.
And that’s where our own research comes in: the evolvable-agent architecture, to a point, configures an overlay network to take advantage of all its resources but it’s not there yet: it would be necessary to include more self-star properties such as self-management and self-configuration, so that the user should have to include just performance targets (obtain solution in a certain amount of time), and the system would set its parameters itself to attain it. A step in the right direction, anyways.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s