Jan 19, 2006

Wow, that’s a mouthful! And what the heck does it mean?

Basically, we are talking about computing systems that are smart enough to 1) know what’s wrong with themselves when a problem happens, and 2) actually take action to fix the problem on the fly. Results are coming in that say this type of technology reduces the time required to fix a problem by 90%! Now, that’s smart! And IBM is at the forefront of the push…

Read all about it on their Autonomic Computing Blog. Neat stuff.

This goes hand in hand with On Demand initiatives – a buzzword for the last few years in the IT management industry. IBM calls it On Demand Business, Computer Associates calls it their On Demand Computing Initiative. A main part of on demand business comes down to faster response to high demand for certain IT resources. If application A is suffering from performance problems, On Demand calls for automatic redistribution of resources to compensate, either from a reserve of capacity or ideally from unused resources that are currently “mis-allocated” to application B that isn’t being using at the moment. Think of it as resource allocation based on dynamic need, rather than static allocation based on a peak usage analysis.

For years, IT architecture has been designed to handle peak instantaneous arrival rates rather than building for average usage. The result is usually large scale “over building” to compensate for peak usage, which means that a lot of resources lay dormant during non-peak periods. As an easy example, think of a classic billing system (which runs all night) and a corporate intranet site (which is accessed all day). With On Demand computing, these two applications share IT resources (CPU’s, bandwidth, etc) resulting in a dramatic reduction of overall cost by driving up the average utilization of each resource across the board. Don’t double build – build it with On Demand technology. Self-Managing Autonomic Computing takes this one step further by bringing automatic problem resolution to an on demand computing environment.