While large enterprises have detailed disaster recovery plans in place which often include pre-configured and mirrored backup file-servers, most small to medium enterprises cannot justify the investment required to implement disaster recovery plans of this nature and must rely solely on the integrity and recoverability of their backups.
In the extreme case, a loss of data can destroy a company entirely. In reality, experience shows that the vast majority of companies have one or more of the following elements present:
Ensure an authorised person is accountable and that this person checks the log files on a frequent basis.
David Lees, director at End2End Business Solutions
* Data is backed up infrequently and irregularly. The danger of this is that, assuming the data backups are recoverable, the actual data is obsolete.
* Data backups entail copying data from one hard drive to another, or from one PC to another. The danger of this is that, in the case of theft, fire or hard drive failures, all of the data copies will be lost.
* Data is written to CD, DVD or flash drives. This method of backing up has no inbuilt integrity or verification checks, thus there is no guarantee that the data is recoverable. In addition, in many instances, the data required to be copied does not fit onto a single CD and must be spanned across multiple CDs. This further adds to the risk of data not being recoverable.
* A procedure is in place whereby vital data is written to a removable hard drive and taken off-site daily. While this may seem to be an efficient solution, my experience is that the data backup files are overwritten every day or every alternate day. Thus it's easy to overwrite "good data" with corrupt data before the corruption becomes apparent. This can easily create an environment where all data backups contain corruptions that render vital data unusable.
* A well-implemented backup procedure is in place whereby a series of backups are created and overwritten sequentially in a way that minimises the possibility of data loss. However, the backup media (tapes, CDs, etc) are used in perpetuity and are never replaced with new media. Data tapes and CDs have a finite useful life and need to be replaced to ensure the backup integrity remains valid.
* One or more of the above backup procedures is in place; however, the log files generated by the backup process are not routinely checked to ensure the backup process is completed without errors. (Or, the process does not generate any output that confirms the correctness of the process.) A data backup could be defined as a file or multiple files that can be restored. Clearly if the backup process terminates improperly or incompletely, the resultant output will not conform to this definition.
I suggest the following backup procedures be put in place:
* Implement a backup process that does generate a report that indicates the success or otherwise of the process. This immediately precludes a manual process whereby an individual is tasked with writing data to flash disk or CD on an infrequent basis.
* Ensure an authorised person is accountable and that this person checks the log files on a frequent basis.
* The backup procedure should include the following:
- A series of at least three daily or weekly backups that are overwritten sequentially every three days/weeks. The frequency of the backups should be determined by transactional volume, ie the more transactions that are being processed, the more frequently you need to backup.
- A monthly backup should be stored off-site for between six and 12 months (or longer depending in compliance requirements). Thus, at any point in time, a series of daily or weekly and monthly backups exist. This creates an environment whereby it's always possible to recover data whether it be a day old, a week old, a month old or older. The chance of multiple backups all being irrecoverable is highly unlikely.
While it may seem expensive to implement a sound backup policy due to the cost of the media requirements, the initial cost is likely to be a fraction of the expense and lost revenue that may be incurred in the event of a data loss where no recoverable backups exist.
Share