Backup

An organisation will be fully dependent on its Information System. Failure or loss of data could easily lead to total failure of the business. An important aspect of the organisation's IT policy will deal with provision for backup. A backup strategy will identify what data is to be backed up, where the backup copies are to be stored, how often or under what circumstances backup should occur and what procedures must be adopted to restore lost data in the case of system failure.

The purpose of backup is to cope with the situation when one or more aspects of system security have failed. Thus backup may be as simple as allowing a file that has been accidentally overwritten to be recovered or it may be designed to cope with the complete destruction of a computer installation. In general the greater the degree of backup the greater will be its cost. An organisation has therefore to balance the costs of maintaining backup against the potential cost of not having it when needed.

A backup system consists of two phases. The first is a set of routine procedures carried out when the computer system is working normally. These maintain data that will allow phase two to be implemented when needed. Phase two of a backup system is a set of procedures used when the system security has failed in some respect. Phase two uses the routine data collected during the first phase to reconstruct the damaged part of the system. In some critical applications it may be essential for phase two to be implemented to restore a working system immediately, In other situations it may be sufficient that the damaged element can be restored over a longer time scale.

The method of backup depends to a large extent on the nature of the data that is being backed up. If the data never or rarely changes as is the case with applications software, then it would be sufficient to keep copies on tape or disc in a safe place - for example in a fire proof safe, possibly in a separate building. Any occasional updates or changes could be copied and similarly stored.

In the case of batch processing systems, where a transaction file is merged with a master file to produce a new master file then the generations method would be used. Here the new master file is called the son. The previous master file is called the father. The father generation master and its transaction file are kept as backup. When another transaction file is merged with the son, then the son becomes the father and the father becomes the grandfather. Three generations of file are normally kept together with their associated transaction files If the current (son) master file is destroyed it can be re-created by merging the father generation master and transaction files again.

Greater problems occur with transaction processing where changes to records are made randomly as they occur. If access to the data is critical then the file may be mirrored. This is when an exact copy of the file is maintained on a different disk. Both files are changed together. If the disk fails, corrupting or making the file unavailable then processing can continue using the image. This protects against loss of the file due to disk failure. It does not however allow recovery from situations where invalid data has been processed or records accidentally deleted. Mirroring would normally be used in conjunction with a regular backup procedure as described below.

Random access files can be backed up by regularly copying the entire file to a tape or to another disk. Special tape devices called tape streamers are available for this. The frequency of copying would depend on how active the file was. This in itself does not allow the file to be recovered since the copy is a snapshot of the file and the data will have changed since the copy was made. In addition to the copy a transaction log must be maintained. This is simply a file containing details of every transaction made since the copy was done. If the file is lost or corrupted then it can be restored by merging the transaction log with the copy to reproduce the original.

©LEV