The entire workload of a system can be broken down into discreet “workloads.” Managing inbound web traffic might constitute one type of workload; managing the ecommerce component of your site may be another; processing payroll could be a third, and producing quarterly sales reports could be a fourth.
At the end of the day, regardless of the specific task that the workload accomplishes, it will generally consist of four components: an application; a database of some sort; an application server; and an operating system.
Now, for efficiency’s sake, there can be a benefit in creating individual systems dedicated to a particular workload. 
This was the idea behind hardware and now software appliances which bundle application, database, application server, and JeOS into a single image.
The appeal especially of such virtual appliances is that in addition to improving workload performance, they make workloads portable. Functioning as a self-contained unit, such a workload can run anywhere. Being able to “run anywhere” is a blessing and a curse. The blessing part is the efficiency and flexibility you gain. The curse is that, in order to safely run anywhere, you need to figure out how to manage security, access privileges, log reporting, etc., when “anywhere” turns out to be somewhere outside your control.
The way to lift the curse is to make workloads “intelligent.” For our purposes, you can consider a workload “intelligent” when it knows where it can and cannot run (in the cloud, for example), when it determines who can and who cannot access it, and when it monitors all activity and generates reports on when it was accessed by whom and what happened.
Traditionally, intelligence resides outside the workload in the humans managing system security and the like. The next stage of workload evolution demands that intelligence migrate into the workload itself.
Posted
Dec 02 2009, 10:06 AM
by
maxled77