Internal Info: Admin Computing Personnel Safety Stores Teaching Web email Links: Funding Bodies Learned Societies University of Oxford World Wide

Computing  Data Storage and Backup

What to Store

There are in general two types of item stored on computers: Programs and Data. Generally, programs should not be considered for backup purposes except in the case of making a direct copy of the installation media in case the originals fail; most modern programs will not work if simply copied from one machine to another. Some programs may have configuration files holding your program settings, which you may want to keep.

Data will include items such as letters, diagrams, experimental data, program source code, and anything else created by a computer user. It is assumed that the user would be at the very least inconvenienced if their work was lost forever. Therefore, to state the obvious, frequent backing up of data is of great importance.

A Backup Method

When working with data a suggested strategy is to work with three copies:

• Your working copy which you save frequently.
• A copy on a different disk, for example a network drive. Copy your working copy to this daily or half daily.
• A day old or half-day old copy. If you made a mess with your working copy and its backup, you can return to this one.

Archive and Backup

There is a distinct difference between archiving data and backing it up, and this is reflected in the different hardware and software used to implement each.

Archive

Long-term storage; keeping a complete set of data for historical reasons.

Backup

Short to medium term storage; storing those parts of a set of data that have changed, for disaster recovery reasons.

Archiving

The usual sort of thing that should be archived is something like a book or thesis. The media on which the original document was stored during its composition may have a short lifetime. Either it will fail, or someone will erase it. There is no compulsion on other people to look after any of your documents, especially if you are no longer part of their group.

Archive media should either have a long shelf life, or more than one copy can be stored on less reliable media.

Examples of Archive Systems

• CD-ROM and DVD-ROM. Standard media which are likely to be readable on future systems. They cannot be erased by magnetic fields and are tolerant of dust. However, damage can occur by severe scratching, and the incorrect use of CD writing software. Use good quality media, and avoid the use of certain inks which may destroy the disk.
• Paper. If the data lends itself to being printed on paper, this is a reliable method if stored correctly. Paper takes up a lot of space, is likely to deteriorate and can be damaged by water and fire. Further editing by computer is most inconvenient, requiring scanning and proof reading of the scanned results.
• Heirarchical File Server (HFS). The University's HFS system can archive data as well as providing for backup, but special permission needs to be granted.
• Multiple "less reliable systems", such as two sets of floppy disks and a hard disk. Especially effective if the systems are different and stored in different places. The danger is that all the systems may fail over time.

Backup

The fundamental 'backup' is frequent saving. Do not work on a document for hours before saving it - software, hardware, and mains supplies are not that reliable! Save and name your document at the time of its creation, then click on the Save button every five minutes or so. The exact time between clicks of the Save button depends on how much you like retyping!

The word "backup" earlier on in the previous paragraph is in inverted commas because this does not really describe a backup in the true sense; if your machine was to be switched off, you would not have a secondary copy. A danger of frequent saving with no backup is that you save a new version over the old each time. If your latest version of the document was corrupted it would overwrite the good version on disk.

Examples of Backup Systems

• Floppy disks. These ubiquitous items hold 1.44MB of data, but they do not have a very long life. They are highly susceptible to dust and magnetic fields. Data stored on floppy disk may last for five years, but a disk may fail within months if stored in the wrong environment or up to ten years if looked after.
• Secondary hard disks. A physically separate hard disk from your main working hard disk, either an internal disk or an external disk in an enclosure connected by USB or firewire. Much more reliable than floppy disks or zip disks as they are sealed from the environment. They mainly fail due to being subject to mechanical shock, especially when they are in use.
• Zip drives and other removable disks. Good for transporting and storing larger amounts of data. Reliability similar to floppy disks, or perhaps a bit better. For large documents and pictures, and for moving data between home and work.
• Memory sticks. Not really designed for backup, except for the very short term. Can be easily erased, and are mainly useful for moving data between home and work.
• Network shares e.g. "the P: drive". The advantage of using network drives it that the drive is on a completely separate machine, perhaps in another room, reducing the points of failure for your data. Also the server serving the network share will usually have nightly backup giving a third failsafe for your data.
• Tape backup systems. Usually attached to servers, although they are available for virtually any mainstream computer. They usually require, and come with, specialised backup software. The computer in question is left on and the software will schedule automatic backup. This will be one of two kinds: full and incremental, where the incremental backup saves only files that have changed since the last backup. Tape backup should be used in addition to the normal safeguards of saving frequently during the day. A backup strategy should normally be employed to rotate tapes and provide a selection of data from previous days and weeks. See the section below.
• Heirarchical File Server (HFS). The University's HFS system will provide automatic weekly backup for those systems which are registered with it. If a good case is made, automatic backup can occur every night. Manual backups can, in any case, be made at any time using the software supplied. The 32-bit Windows software is available for installation from the Physiology file server.

A Backup Strategy

An example of simple backing up using a single computer is:
Keep your working copy of a document on a computer's hard disk. Frequently save it, while less frequently making a copy onto other media (not another directory or partition on the same hard disk). You may also wish to make an even less frequent copy with a different name in case you wish to return to it. For example, if I am going to make changes to a working program or printed document, I will make sure I have a copy of the 'released' document before starting my major changes to the next revision. This sort of thing can be achieved automatically by using revision control software - Unix users have "RCS" available to them.

An example of a professional backup strategy:
Computer users should look after their own data by saving and backing up frequently during the day. Do nightly incremental backups: Use a tape drive or similar system, and buy a set of media from a reputable manufacturer. Rotate your tapes or disks nightly. An example system using nine tapes would be: