Backups |
No-one likes losing data that they have entered into a computer. Lost data represents a waste of time and effort. For companies the loss of data can be especially serious. Many companies store essential information such as customer accounts or stock databases on computer. Loss or corruption of this information could cost a company a lot of money.
If data is very important then appropriate security measures should be put in place to try and avoid any data loss or corruption. However problems can still occur :
Therefore extra copies of important data should be made on a regular basis. These copies are known as backups. If the original files become corrupt then the data can be restored from a backup copy. A company should have a backup strategy which sets out how backups will be made.
Often the method used to make backups will depend upon the type of processing that is being carried out :
| Backups by Type of Processing | |||
| Batch Processing | |
Backups are kept using the Ancestral or Grandfather-Father-Son system. | |
| Transaction Processing | |
Periodically a copy of the master file called a file dump is made. Recent transactions are stored in a transaction log file. | |
| Real-time Processing | |
A lot of backup hardware is employed to reduce the risk of failure. | |
A backup is not the same thing as an archive. An archive is a file containing old data which is no longer used but which must be kept. Archive files are often kept under the same conditions as backups.
.
A backup strategy is a plan which details how backups will be made and how data will be restored in the event of a system failure. There are many important factors that should be considered in a backup strategy. These include :
How often should data be backup up ?
If a file is corrupted and you restore the file from a backup copy then some data may still be lost. This is the data that was added to the file since the backup was made. If backups are made once a week then a whole week's work could be lost. On the other hand if backups are made once a day then only a day's work would be lost. Backups should therefore be made as regularly as possible.
Unfortunately backups cost money to produce. Storage media must be purchased and staff time is taken up. Therefore a balance has to be struck between making backups very frequently and keeping backup costs down.
As a general rule files which change frequently should be backed up often whereas files which do not change very much can be backed up less frequently.
When should data be backed up ?
Backing up data can tie up a computer so that it can not be used whilst the backup is in progress. Therefore backups are usually made overnight or at weekends when a computer system is less likely to be in use.
What storage medium should the backups be stored on ?
There are a variety of different storage media that backups can be stored on. These include magnetic and optical disks and magnetic tapes. The best medium to use in a particular situation will probably depend upon the volume of data to be backed up and the speed at which the backup must be performed.
Where should backups be stored ?
There is no point in making a backup copy if the backup may be damaged. These guidelines should be followed to ensure that backups are stored safely :
Should backed up data be compressed ?
Data that is backed up can be compressed to reduce the amount of storage space that it will require. Compressed backed up data must be decompressed before it can be restored. Therefore backups are usually only compressed if they are unlikely to be accessed regularly.
Who will carry out the backup procedure ?
In a small company each employee may make his or her own backups. In a larger organisation however the task of producing backups will usually be performed by one selected individual. This individual can be properly trained in using the backup procedure and can keep accurate records of the backups that have been made.
What will happen if data is lost ?
If data is lost then it will have to be restored from the backup. A disaster recovery plan should be in place which describes how the data will be restored.
How many copies to keep ?
It is a good idea to make more than one backup copy. This gives you the extra assurance that if a file is corrupt and the backup is also corrupt then you have another backup which may still be operational.
In a batch processing system the File Generation, Ancestral or Grandfather-Father-Son backup method is used. Remember that in a batch processing system all transactions are entered into a transaction file. This file is then used to update the current master file to produce a new updated master file.
| When processing takes place the master file is known as the
father file. Processing the father file with the transaction file
produces the new master file which is known as the son file. The father file and transaction file are stored securely. If the son file becomes corrupt the father file can be processed with the transaction file again to reproduce the son file. In this way the father file and its associated transaction file act as a backup of the son file. An older copy of the master file called the grandfather file is also kept. If the son file and the father file are both corrupted then :
Every time processing occurs the tapes are moved back one generation. So the father file becomes the grandfather file and the son file becomes the father file. The old grandfather file is erased to be reused. |
 |
Backups for a transaction processing system are produced in a different way to those for a batch processing system. At the end of each time period a backup copy of the master file is made. The time period between each backup could be a day or a week. Making a complete copy of the master file like this is known as file dumping. The copy of the master file is known as a file dump.
Every time an alteration is made to the master file, since the last backup was made, details of the alteration are stored in a transaction log file.
If the system breaks down then two things happen :
By following this process no data should be lost. If there was no transaction log file then all of the transactions since the last file dump would be lost.
The consequences of failure in many real-time systems (e.g. flying a missile/passenger plane, running a nuclear power station) are immense. There is not a lot of point in keeping backup copies of data in a real-time system as there is unlikely to be time to restore backups if anything goes wrong. When a real-time system is developed the emphasis is placed on ensuring that the system will not go wrong rather than on designing a backup method to restore the system if it does go wrong.
Therefore most real-time systems incorporate a lot of spare hardware. Having much more hardware than you need to do a job means that if some hardware fails you can carry on doing the job. For example three identical computers may be used to decide which direction a missile should fly in. The computers each have a "vote" on which direction the missile should fly in. Because the computers are identical all three will normally agree on which direction to take. If one computer goes wrong then it will be outvoted by the two working computers so the missile will still function properly despite the failure. If two computers go wrong you are in trouble.
Sometimes backing up data to disk can be useful in real-time systems. Although this data could not be used to put the system back into its proper state, it could be used later to analyse why a system failed. The "black box" that records flight data in an aircraft is an example of this type of system.
GCSE ICT Companion 04 - (C) P Meakin 2004