Overview
The statement that system analysts should be in charge of repairing and controlling system problems when they occur cannot be more misinformed. Taking care of a system problem by repair or replacement of faulty parts only returns the system to the previous condition and does not assure that the problem would not recur. So, the objective of remedy should be fault prevention.
Forms of prevention
When a system does not behave in accordance with the specifications to which it was built, it is considered to be experiencing fault. The two fundamental approaches to constructing a reliable system are fault tolerance and fault prevention. (Tamer M. O.,1999). While fault tolerance measures constitute design time features to be built in system which detect and remove faults that can result in system failure; fault prevention is the system analysts’ task to ensure a reliable system after it has been implemented and is under use.
Analyze risks involved. Identify the probable points of failure of the system and how far reaching could be the impact on business. Take steps to ensure maintain business continuity if the system will exhibit faults, and implement monitoring systems for critical components such as operating system performance, database stability monitoring and hardware maintenance.
Ensure safe transactions. Incorrect input data or database deadlocks can lead to system failures. While data validation is strictly recommended at client level, database design should implement transaction abort on fault detection.
Maintain redundancy. Hardware and software failures are the most common forms of system failures. Keeping a second set of servers for load balancing and a database replication server can make the system scalable to take on additional load and also act as buffer for business continuity if a fault is detected in primary system.
Identify symptoms of failure. An operating system running continuously leaks memory and needs restarting; scheduled restart actions can prevent future failures. Look for garbage accumulation and corruption of data and take preventive steps on detection. In a web system, adapt to circumstances for predicting excessive connection requests and manage ahead in time.