Exit strategy
One of the main enemies of digital preservation are software "solutions" that claim to be the last digital preservation platform you'll ever need, and won't make it easy (or even possible) to export your own data when you want to migrate them to a different platform. One should run away from those products as fast as possible. A good preservation platform should offer a transparent, documented, and cost-free process to export all the data into a format that is readable by other platforms, so that one can walk away from it at any time, no questions asked. In other words, archival software must allow the plannig of an exit strategy.
Pocket Archive has backup and restore tools accessible via its command-line interface. These tools can be used both to create replicas (as described above) and to migrate the data to a different platform. The exported metadata are in RDF, which is an open format widely supported by many programming languages, so that one can write scripts to transform the RDF metadata dumps into a format suitable to any other system. Files are transparently copied to disk, but they don't have the original layout as they were submitted in the SIP. The metadata contain their original and archival location, and here again some logic handling is necessary to identify the files, relate them to their metadata, and move them to another location.
These exports can be also used as replicas (as long as they are moved to a separate support, as explained earlier in the 3-2-1 replication guidelines). These replicas are more transparent as they don't need Pocket Archive to be read and understood, but they take longer to generate and may be larger (however only the database has to be exported — files always keep their layout).
The backup tool can optionally export the full application and content model configuration, so that a whole archive can be duplicated on a different location, or to provide administrators and developers with the information needed to migrate the archive's content model to a new system.
The following steps allow for a full-circle back up and restore of a complete archive. These steps assume that Pocket Archive is installed in a Linux system.
- Back up the full archive via the command line tool:
pkar backup <destination dir>(where<destination dir>is the parent directory of where the backup will be created. if the folder does not exist, it will be created as long as the user running the command has the permissions to do that. The archive will be created in a new directory identified by the current date & time. The default options create a full back up of data files, aka opaque resources (in theoresdirectory), descriptive resources serialized as RDF/Turtle (thedres.ttlfile), and the current configuration (in theconfigdirectory). These copies are done by verifying all the checksums of the copied data files. - Take all necesary steps to safeguard the generated backup according to your preservation plans, for example replicatiing them to on-site and/or off-site locations (see Replication).
- It can become necessary to restore the archive from a backup in case of disaster, to migrate the data to a different platform, or just for routinely testing the validity of the backup.
- For the first scenario, the Pocket Archive command line is used again.
PKAR_CONFIG_DIR=<current config> pkar restore <source directory>restores the full archive from the archive directory indicated by<source directory>to the location set up in<current config>. The location of the configuration, which contains pointers to the opaque and descriptive resources' locations, can be set with thePKAR_CONFIG_DIRenvironment variable as illustrated in the example. - The
app.luaconfiguration may need to be reviewed and adjusted to ensure that the opaque resource and descriptive resource DB paths point to the new locations. - The archive can be run by specifying the new configuration location in the
PKAR_CONFIG_DIRvariable, e.g.,export PAKR_CONFIG_DIR=<new config location>followed by anypkarcommands.
In future versions, Pocket Archive will provide simplified tools for backing up and restoring archives, and for better fixity support.