Backup Strategy ¶
Backups are usually in the responsibility of the system administrator. Creating backups obviously does not improve the security of a TYPO3 site but they quickly become incredibly useful when you need to restore a website after your site has been compromised or in the case of a data loss.
Components included in the backups ¶
To restore a TYPO3 project you need to have a backup of at least the following data directories:
The directory structure is not
fixed: Especially with
based instances, a separation of the
web server document root
can be achieved.
This is recommended from a security point of view since critical parts of
the TYPO3 instance can be located outside of the web document root and are then not
directly accessible from the outside. Important parts are the
for details. The “config path” should
be included in backups, while the “var path” could be left out since its content
will be - with most instance configurations - recreated automatically if needed.
You do not need a backup of the
directory, due to the
fact that all files are re-generated automatically if they do not
exist. Also a backup of the TYPO3 source code is not needed (unless
changes were made to the source code, which is not recommended). You
can always download the TYPO3 source packages from the TYPO3 website,
even for older versions of TYPO3.
In addition to the data directories listed above, a backup of the
database is required. For MySQL the command line tool
for Microsoft Windows) is a good way to export the
content of the database to a file without any manual interaction (e.g.
as an automated, scheduled system task).
Once a backup has been created, it should be verified that it is complete and can be restored successfully. A good test is to restore a backup of a TYPO3 project to a different server and then check the site for any errors or missing data. In a perfect world, these restore checks should be tested frequently to ensure that the concept works and continues working over a time period. The worst case would be that you rely on your backup concept and when you need to restore a backup you notice that the concept has not worked for months.
Time plan and retention time ¶
In most cases you should create a backup once a day, typically at a time when the server load is low. Rather than overwriting the backup from the previous day you should create a new backup and delete older copies from time to time. Just having a backup from last night is not sufficient for recovery since it would require that you notice the need for a restore within 24 hours. Here is an example for a good backup strategy:
- keep one daily backup for each of the last 7 days
- keep one weekly backup for each of the last 4 weeks
- keep one monthly backup for each of the last 6 months
- keep one yearly backup for each year
Backup location ¶
Backups are typically created on the same server as the TYPO3 instance and often stored there as well. In this case, the backup files should be copied to external systems to prevent data loss from a hardware failure. If backups are only stored on the local system and an attacker gains full control over the server, he might delete or tamper with the backup files. Protecting the external systems against any access from the TYPO3 server is also highly recommended, so you should consider “fetching” the backups from the TYPO3 system instead of “pushing” them to the backup system.
When external systems are used they should be physically separated from the production server in order to prevent data loss due to fire, flooding, etc.
Please read the terms and conditions for your contract with the hosting provider carefully. Typically the customer is responsible for the backup, not the provider. Even if the provider offers a backup, there may be no guarantee that the backup will be available. Therefore it is good practice to transfer backups to external servers in regular intervals.
In case you are also storing backups on the production server, make sure that they are placed outside of the root directory of your website and cannot be accessed with a browser. Otherwise everybody could simply download your backups, including sensitive data, such as passwords (not revealing the URL is not a sufficient measure from a security perspective).
Further considerations ¶
More sophisticated backup strategies, such as incremental backups and distributed backups over several servers, geographically separated and rotating backups, etc. are also achievable but out of scope of this document.
Due to the fact that website backups contain sensitive information (backend user details, passwords, sometimes customer details, etc.) it is highly recommended to consider the secure encryption for these files.