General

Data safety and backups nowadays are of enormous importance in every company – especially in data processing-intensive branches. The data of a company equals working time and thus must be handled like an asset or a finished product with an according property value.

But in order to calculate the (proprietary) value of a document or a whole data base, other things, too, have to be considered besides the invested working time: In case of data loss, even missed sales, image loss and, possibly, dead times (with lost sales and fines in other company locations) have to be calculated as well. A long-time study shows that 60 % of the companies that suffered a total data loss have not been able to cope with the resulting financial difficulties and had to give up within one year.

The probability of a partial or total data loss on the one hand depends on the used hardware that makes up the technical environment of the data processor. On the other hand, the security measures that have been taken against viruses and hackers are very important. The main cause for data loss, however, still is human faulty handling, which is responsible for almost 70 % of such total breakdowns.

That is why every company that has data which gained a certain value, should think about a backup system. The most important decisions regarding successful data security have to be made in advance in order to achieve the proper strategic focus: It should be considered which data are to be saved, which system is the most effective solution for your requirements and, finally, which storage media should be chosen.

The decision about the data to save has to take into consideration the following three points in particular: the user's habits, the software equipment/configuration of the company and the size/volume of the backup server. If you just want to save specific user-data, only a small backup server is needed. The problem here: If the hard disc fails, all the software has to be reinstalled. This shows that all considerations have their certain influence on the backup strategy.

Mostly, various backup strategies are deployed in combination in order to link all the specific advantages of the selected methods. Three general kinds of backups have to be differentiated: the image, the full backup and various incremental backups. An image is an on-to-on copy of a storage media. Unlike the full backup, it also includes file allocation tables, boot sector and other system-relevant data in the image. The incremental backups are different especially in temporal reference. This means that it is important to think about which older backup the ongoing data are harmonized with, and if it is a version backup or if the old data should simply be overwritten.

The decision about the backup media is particularly influenced by the required storage time of the data. Storage media also succumb to material decay, which should be taken into account. Hard disc drives, ZIP- and JAZZ-devices guarantee shorter data-durability than CD, DVD and, with suitable storage, also tape drives. With storage times of several decades the storage of technology gets important, too. Otherwise viable data could possibly not be read and interpreted any more.

Finally the configuration of the backup infrastructure should briefly be mentioned. From backup by CD-burning or RAID-systems on local computers up to redundant server solutions everything is possible today. You should choose a system which is optimized for your company's data processing structure.

The Appropriate Backup Strategy

In the following, we try to give an insight into the fundamental terms and techniques of data security as well as some details for the optimal backup strategy of your choice. You should know that, regardless of your backup choice, you CANNOT save files which are being accessed at the time of the backup. Usually you can see corresponding messages in the log-file, showing what data was not saved.

Why Generations?

At the time you actually need a backup, the hard disc drive is rarely completely destroyed. Oftentimes only an inadvertently deleted or overwritten file or a corrupted file is needed. Normally one notices this loss only days or even weeks later and during the last backup the corrupt file has been saved. Sometimes you'd even want to access an older version of your file. In such cases the generations principle has proven valuable. If you work with removable media, you save different generations to different media. If you just save to another drive, you can use the inbuilt "generations" function. In the following, two examples of different backup strategies are shown.

Backup Types

The various backup types initially are not connected to the generations. With these backup types we only decide, WHAT (in which scope) is saved. You should first comprehend the following terms and only afterwards decide how you backup.

• Complete Backup
  
  This is the easiest method: all files, no matter if they were changed since the last backup, are saved. With this method you at all times have hold of an up-to-date image of your data. Disadvantage: Every backup requires a lot of space on the target media and the expenditure of time is rather high for big databases.


• Incremental Backup
  
  With this method only those data are saved, which have been changed since the last backup or which have been added afterwards. Thus incremental backups are done quickly and need little space. Incremental backups are executed in an alternating rhythm with complete backups. For example: complete backup on the weekend followed by daily incremental backups. So on Monday evening only those data are saved that have been changed on Monday, on Tuesday evening only those changed on Tuesday and so forth.


• Differential Backup
  
  This method is very similar to the incremental backup and is also used alternating with complete backups. But here, staying with the above example, on Monday evening those data that have been changed on Monday are saved, on Tuesday evening the data changed on Monday and Tuesday are saved and so forth, thus being all data after the last complete backup. Advantage: You only need the complete backup and the last differential backup to restore the original data. With an incremental backup on Thursday you would need the complete backup as well as the incremental backups from Monday to Thursday.

If you have the facilities you should at frequent intervals also outsource your backups to removable storage and keep them save in other places.

Examples for Backup Strategies

It is advisable to create several smaller backup profiles and not just always save a whole drive at once. Smaller backup profiles are a lot faster to finish and don't interfere so much with the work in progress. We present here two examples for "lean" backups with various generations:

• Example 1:
  
  Usage of removable storage devices, compressed backups, usage of complete backups and differential backups applying the grandfather-father-child-principle: For this model you'll need 21 external storage devices in the best case: for each month one media ("January" to "December"), 5 week media ("Friday 1" to "Friday 5") and 4 media "Monday" to "Thursday" for the weekdays. The backup starts with a month backup - complete backup of the data - this is the "grandfather". Additionally every Friday complete backups are made - that is the "father". From Monday to Thursday differential backups are created - the "children".
  
  With this method you have access to various generations: all changes during the week (media "Monday" to "Thursday"). For the data of last month: weekly copies (media "Friday 1" to "Friday 5"). For last year's data: monthly copies (media "January" to "December"). So you have access to an extended period of different versions of your data. If you do not operate with large amounts of data you can possibly pool media. Of course you can adapt this variant to your needs and store quarterly instead of monthly backups. So you need less media, but you also have less generations.


• Example 2:
  
  Usage of a second hard disc drive (backup directory), uncompressed backups, one-time complete backup, incremental backups, usage of internal "generations" function: You define a "backup-directory" on your second hard disc drive. With this version in principle you only need one complete backup. You define a second profile with the same settings but as incremental backup, i.e. with 10 generations. Whenever a file is changed the system first checks whether this file already exists within the backup. If so, the file is moved to a sub-folder named "generation 1", and only afterwards the original file is copied to the backup folder. If this same file is saved on the next day, too, because it has been changed again, the copy from "generation 1" is moved to "generation 2", the copy from the backup-folder is moved again to "generation 1" and a new copy of the original file is again saved to the backup folder.
  
  With this method you always find the most recent version of your data in the backup folder. The sub-folder "generation 1" contains the second-to-last version etc. In using incremental backup only additional copies of the changed data are generated. Therefore backup time and needed space are relatively small. This variant of "generations" doesn't work for compressed backups.

Additionally it is advisable here as well, to also save backups to removable storage devices, as these can also be stored in other places.

These data you should save as well:

You should always save your "operational" data. They are your "capital", normally a lot of work has been put into them, loss of this data is oftentimes irrecoverable.

To keep backup expenditure - also regarding necessary storage requirements - within reasonable limits, you should think twice which data are worth saving for you. When saving your data don't forget to also save important data not stored in your data directories.

Pay attention to also save the following files:

• style sheets and macros of your word processor
• bookmarks of your browser
• registry
• system data
• INI-files
• address books

Make sure that your backup system can communicate with all your drives with read/write-access. These usually are all hard disc drives reachable by your operating system, no matter if formatted with FAT, FAT32 or NTFS, disc drives, removable storage devices such as ZIP or JAZZ drives as well as accessible drives within your network.

Moreover it also should work together with CD-recordable or CD-re-writable drives, if corresponding drivers are installed (in connection with UDF-burning-software - PacketCD, DirectCD, InCD). Those drivers also enable access to read/write processes just like an ordinary access to a hard disc drive. Adaptec DirectCD for example is such a driver.

There are also tape drives with corresponding drivers so the tape can be accessed like an ordinary drive.

Notes on Compressed Backups

As saving and creating of large compressed files is rather laborious regarding resources, we strongly recommend to create smaller backup packages with 10 to 100 MB max. 3 reasons for that recommendation:

• Security
  Especially when single files in ZIP-files are changed, defective ZIP-files can emerge. Smaller ZIP-files reduce the risk of losing large data quantity.
• Efficiency and Performance
  To add further files to existing large ZIP-files or to change certain files increases the necessary time proportionally to the size of the ZIP-file.
• Disc Space
  When you update data in ZIP-files, first a copy of the complete ZIP-file is produced, only then the changes are made. This copy is normally saved to the "temporary" directory. Therefore, to make this change, corresponding space has to be provided on the drive containing the temporary directory. Also consider the effort for copying a large file.

  Top