LogiCL
Services Informatiques

LogiCL It Services

Backup



- Data backup cycle -

Full, incremental and differential backups.

All backup schemes are based on full, incremental and differential backup methods. The term "scheme" in fact denotes the algorithm of applying these methods plus the algorithm of the archive cleanup.

Comparing backup methods with each other does not make much sense because the methods work as a team in a backup scheme. Each method should play its specific role according to its advantages. A competent backup scheme will benefit from the advantages of all backup methods and lessen the influence of all the methods’ shortcomings. For example, weekly differential backup facilitates archive cleanup because it can be easily deleted along with the weekly set of daily incremental backups depending on it.

Backing up with the full, incremental or differential backup method results in a backup of the corresponding type.


Full backup

A full backup stores all data selected for backup. A full backup underlies any archive and forms the base for incremental and differential backups. An archive can contain multiple full backups or consist of only full backups. A full backup is self-sufficient - you do not need access to any other backup to recover data from a full backup. It is widely accepted that a full backup is the slowest to do but the fastest to restore.

A full backup is most useful when:

-you need to roll back the system to its initial state
this initial state does not change often, so there is no need for regular backup.

Example: An Internet cafe, school or university lab where the administrator often undoes changes made by the students or guests but rarely updates the reference backup (in fact, after installing software updates only). The backup time is not crucial in this case and the recovery time will be minimal when recovering the systems from the full backup. The administrator can have several copies of the full backup for additional reliability.


Incremental backup

An incremental backup stores changes to the data against the latest backup. You need access to other backups from the same archive to recover data from an incremental backup.

An incremental backup is most useful when:

-you need the possibility to roll back to any one of multiple saved states the data changes tend to be small as compared to the total data size.

It is widely accepted that incremental backups are less reliable than full ones because if one backup in the "chain" is corrupted, the next ones can no longer be used. However, storing multiple full backups is not an option when you need multiple prior versions of your data, because reliability of an oversized archive is even more questionable.

Example: Backing up a database transaction log.


Differential backup

A differential backup stores changes to the data against the latest full backup. You need access to the corresponding full backup to recover the data from a differential backup. A differential backup is most useful when:

-you are interested in saving only the most recent data state the data changes tend to be small as compared to the total data size.

The typical conclusion is: "differential backups take longer to do and are faster to restore, while incremental ones are quicker to do and take longer to restore." In fact, there is no physical difference between an incremental backup appended to a full backup and a differential backup appended to the same full backup at the same point of time. The above mentioned difference implies creating a differential backup after (or instead of) creating multiple incremental backups.

An incremental or differential backup created after disk defragmentation might be considerably larger than usual because defragmentation changes file locations on the disk and the backup reflects these changes. It is recommended that you re-create a full backup after disk defragmentation.

The following table summarizes the advantages and shortcomings of each backup type as they appear based on common knowledge. In real life, these parameters depend on numerous factors such as the amount, speed and pattern of data changes; the nature of the data, the physical specifications of the devices, the backup/recovery options you set, to name a few. Practice is the best guide to selecting the optimal backup scheme.


ParametersFull backupDifferential backupIncremental backup
Storage space Maximal Medium Minimal
Creation time Maximal Medium Minimal
Recovery time Minimal Medium Maximal



GFS backup scheme

This section covers implementation of the Grandfather-Father-Son (GFS) backup scheme.

With this backup scheme you are not allowed to back up more often than once a day. The scheme enables you to mark out the daily, weekly and monthly cycles in your daily backup schedule and set the retention periods for the daily, monthly and weekly backups. The daily backups are referred to as “sons”; weekly backups are referred to as “fathers”; the longest lived monthly backups are called “grandfathers”.


GFS as a tape rotation scheme

GFS was initially created and is often referred to as a tape rotation scheme. Tape rotation schemes, as such, do not provide automation. They just determine:
how many tapes you need to enable recovery with the desired resolution (time interval between recovery points) and roll-back period
which tapes you should overwrite with the forthcoming backup.

Tape rotation schemes enable you to get by with the minimal number of cartridges and not to be buried in used tapes. A lot of Internet sources describe varieties of the GFS tape rotation scheme. You are free to use any of the varieties when backing up to a locally attached tape device.

Create the backup plan as usual. For the backup destination, choose any storage device where automatic cleanup can be performed, such as an HDD-based storage device or robotic tape library. (Since the space freed on the tape after cleanup cannot be reused until all the tape becomes free, take into account additional considerations when using GFS on a tape library.)

The following is an explanation of the settings that are specific for the GFS backup scheme.

GFS-related settings of the backup plan
Start backup at:
Back up on:
This step creates the total backup schedule, that is, defines all the days you need to back up on.
Assume you select backing up at 8:00 PM on workdays. Here is the total schedule you have defined.

The total schedule.
Schedule: Workdays at 8:00 PM

Weekly/Monthly
This step forms the daily, weekly and monthly cycles in the schedule.

Select a day of the week from the days selected in the previous step. Each 1st, 2nd and 3rd backup created on this day of the week will be considered as a weekly backup. Each 4th backup created on this day of the week will be considered as a monthly backup. Backups created on the other days will be considered as daily backups.

Assume you select Friday for Weekly/Monthly backup. Here is the total schedule marked out according to the selection.

“Inc” stands for the backup that is considered Daily. “Dif” stands for the backup that is considered Weekly. “Ful” stands for the backup that is considered Monthly.

Month 1
Week 1Week 2Week 3Week 4
Su Mo Tue We Thu Fri Sat Su Mo Tue We Thu Fri Sat Su Mo Tue We Thu Fri Sat Su Mo Tue We Thu Fri Sat
- Inc Inc Inc Inc Dif - - Inc Inc Inc Inc Dif - - Inc Inc Inc Inc Dif - - Inc Inc Inc Inc Ful -


Backup:
Full = Monthly
Incremental = Daily
Differential = Weekly

Uses of incremental and differential backups that help save storage space and optimize the cleanup so that consolidation is not needed. In terms of backup methods, weekly backup is differential (Dif), monthly backup is full (Ful) and daily backup is incremental (Inc). The first backup is always full (Ful).

The Weekly/Monthly parameter splits the total schedule into daily, weekly and monthly schedules.
Assume you select Friday for Weekly/Monthly backup.

Important:

Launch the Checkdisk command in order to find and fix logical file system errors, physical errors or bad sectors. To be started before the backup starts.














S
O
C
I
A
L