Let us first look at the traditional troubleshooting methods that are taught.
FORWARD CHAINING.
This method starts at the inputs to the system and works through it sequentially testing each component down the line until the faulty one is found.
Advantages - a logical step by step approach.
Disadvantages - not the optimal method. What if it is the last component in the chain that is faulty?
BACKWARD CHAINING.
Start at the failed output and work your way back towards the inputs of the system until the faulty component it found.
Advantages - a logical step by step approach.
Disadvantages - not optimal. What if the input is faulty?
PROBABILISTIC REASONING.
Look at past failures (historical data) and start testing at the component with the history of most failures.
Advantages - Looks at the probability of failure and checks the items that tend to fail most first.
Disadvantages - Not optimal. What it the item with the highest probability also takes the longest to test and eliminate.
SPLIT HALF REASONING.
The binary search. Start in the middle of the system. If the test there is good, then the first half of the system is ok and you have eliminated half of all the components. Each subsequent test also eliminates half and soon you have isolated the faulty component.
Advantages - Rapidly reduces the number of suspect components
Disadvantages - Does not take probability into account.
ALL THE ABOVE TECHNIQUES COMPLETELY MISS THE MOST IMPORTANT ASPECT - THE BUSINESS BOTTOM LINE.
