The impact of a data loss on a business can be visualized if you consider a manufacturing operation with over a thousand employees, whose pay details are kept in the company's computer system.



Imagine what could happen if these pay details become inaccessible owing to system problems, operator errors or a malicious virus attack. Reconstructing the details is not going to be a short-term operation.

Yet such reconstruction is unavoidable not only to pay the employees correctly but also to comply with regulatory requirements under employment and taxation laws. The cost of such a reconstruction is going to be a major expense item for the company, not to mention the disruption to business operations.

On the other hand, if the company had taken proper actions to make recovery of the lost data possible, the time, business disruption and money costs would be insignificant compared to a complete reconstruction from scratch.

How Data Gets Lost

An awareness of the different ways data could get lost should be the starting step in organizing your defenses. So we look at these first.

Data stored on a computer disk can be lost in several ways.

• User Action



• The computer user deletes a data file or computer program thinking that it is no more needed. It could later turn out that the deletion was premature.



• Accidents



• User deletes file or program, but not with full awareness of what he is doing. For example, a blanket deletion of a large number of files could result in deleting some important files that the user would not have consciously deleted.



• The files or programs were stored on removable storage media like floppy or compact disks or USB drives that were misplaced and could not be found when needed.



• Errors made by network administrators and other administrative persons could lead to the loss of important files or programs that their users might not have deleted.



• System or Infrastructure Problems



• Power could fail before the data could be saved to permanent memory and the user might be unaware that it has not been saved.



• Different kinds of hardware failure, such as contact between the read-write head and the rotating disk platter in a hard disk drive (in normal operation, there is a minute cushion of air between the two), could lead to the data on the disk becoming unreadable.



• Software crashes, as when an application or the operating system suddenly gets stuck halfway through a work session, could result in the work not being saved.



• Data corruption as when the file system or database gets corrupted and the data in the files or database becomes unreadable, or when unintended changes during transmission or retrieval results in erroneous data.



• External Factors



• A natural disaster like an earthquake, flood or tornado destroys the equipment including the data storage media.



• External Interference



• A worm or virus attack results in damage to data



• A hacker intrudes into the system and erases/damages the data



• A thief steals the physical media

Of the above, hardware failure and human errors reportedly account for 75% of the incidents. Data loss through natural disasters is rare. However, if that happens, there is no way to recover the data unless you had stored backups in a physically separate location.

Let us now look at ways to minimize data loss. You can only minimize it; absolute protection even against minor losses is impracticable.

We also look at the possibilities of data recovery if data does indeed get lost.

Organizing Against Data Loss

Backups: You copy all data and programs to a secondary media, preferably removable media that can be stored in a separate location. You can then restore the system or the data from these backup copies in case of a data loss.

While in theory, backup is a foolproof safeguard; in practice things are quite complex. First comes the organizational aspect.

• Taking backups after every change



• Selecting the media to store the backup, having regard to the costs and recovery steps involved



• Keeping track of what data is on what media and which are the latest ones



• Auditing the backup procedures and media to ensure that they are indeed reliable

Numerous options are available for backup.

• Media for storage could be magnetic tapes (least expensive but slow retrieval), hard disks (fast retrieval possible but was comparatively quite expensive until recent times)



• Data could be backed up fully, or incrementally, or continuously, with different cost and recovery implications



• The backed up data could be stored in a way that enables immediate online access or in offsite vaults or disaster recovery centers



• Remote third party backup services could be utilized getting the benefit of their expertise and facilities

The above options are actually only illustrative. There are more options and sub-options available to suit different requirements. The availability of the options and their different implications make selection of a backup policy a complex exercise.

In practice, backup is mostly unsatisfactory in implementation and far from adequately dependable.

RAID: RAID through hardware or software implements redundant storage so that the failure of a single disk does not result in data loss. RAID implementations have different levels with different degrees of safeguards against data loss. RAID is not only a data loss safeguard; it is a means for higher speed data access.

Journaling: An apparently single write operation actually involves several write operations. If some disturbance prevents one of these several operations being completed, the result would be invalid data. One remedy against such an event is maintaining a journal of all changes before actually doing the write operation. If the write is not completed successfully, the operation can be replayed using the journal and completed successfully.

Antivirus: Practically every computer user would have by now become familiar with antivirus software that helps you prevent virus attacks and even recover any damage done by such attacks.

Firewalls: Firewalls act as a barrier between networks of different trust levels. The Internet is a network of no trust level while an internal network has a higher trust level. A firewall could control the data flow between the networks by either allowing only previously allowed network connections or preventing specifically blocked connections. In practice, firewall implementation tends to be inefficient.

User Education: Considering the fact that user errors are a major factor for data loss, proper user education could go a long way in preventing data loss.

Data Recovery

Expert data recovery consultants could recover much of the ?lost? data not only because of their specialized training and experience but also because they have the necessary facilities, such as Class 100 Clean Rooms kept free of dust. Even minute specks of dust could make the thickly packed data on disks unreadable.

These companies use techniques like repairing the file system damage so that the image on the disk could be deciphered as meaningful data, replacing damaged PCBs or read-write heads with matching, healthy ones, removing the platters from the damaged drive and installing them in a healthy drive or a combination of these measures.

It would be apparent that these are procedures requiring high precision, considerable expertise and specialized facilities. Attempting to do it in-house is only likely to make the data completely irrecoverable.