Manipulation and Processing Data

There are two principle modes of operation by which a computer system can process data. They are called batch and transaction processing.

In batch processing, transactions are collected together to form a batch. The batch of collected transactions is then processed in one go. An example of this would be an electricity billing system. A transaction record would consist of a customer identifier together with the current meter reading. Records for all customers whose meters had been read that day would form a batch. Batch processing systems typically require sequential file access.

In transaction processing each record is processed as the data arrives. An example of this might be the system dealing with customer payments where the customer account record is updated as the bill is paid. Since the data does not arrive in any previously known order, transaction processing typically requires random access.

Often the same file will need to be accessed sequentially for some operations (e.g. preparing bills) and randomly (e.g. accepting payments). Since it is difficult to process a random access file sequentially the file would need to have an indexed sequential structure. For the electricity billing system two indexes could be used. One based on post code and house number - to produce bills ready sorted for the post, thus attracting a discounted postal rate, and one based on account number to allow immediate updating when bills were paid.

Interactive processing is a particular case of transaction processing. This type of processing is characterised by immediate update of files combined with user input overlapping with output. This type of processing is often refered to as real-time, or more correctly pseudo real time processing. An example of this would be an airline ticket reservation system where the operator checked the availability of seats on a variety of flights and then made a reservation for the customer for a selected flight. A particular problem with this mode of file processing is to ensure that two operators do not attempt to access the same record at the same time since this could result in selling the same seat twice. This is prevented by the system granting write access to only one user at a time. The record would be locked for other users until the first operator completed his or her transaction. This is pseudo real-time processing rather than true real-time processing because there are often short delays between a transaction occurring and the relevant file being updated. The delays are too short to affect the end user's perception that the system is real-time.

Problems can occur in situations where batch and transaction processing are used together. An example would be a stock control system where data on stock movement was collected together at the end of each day to be processed as a batch while information about stock levels and locations was available in interactive mode. An item could incorrectly appear in stock when the stock file is interrogated while in reality is out of stock and will be shown as such when the day's transactions are processed. The data stored on the system has aged and become less reliable.

© LEV