About Databank Shadowing
Databank shadowing means updating one or more copies of a databank simultaneously.
The master is the `normal' databank file which is accessed for data storage and retrieval. The copies are called shadows. A databank can have more than one shadow.
Any changes to the master databank are automatically made to the shadow, thus protecting data from a disk crash or other event that might cause a databank to be lost.
If a master databank is lost, a shadow will automatically take over from the master and operations can be resumed immediately, assuming the shadow is not also damaged.
A shadow can be transformed to a master databank to permanently replace it and this process is much faster than restoring a databank from a backup copy.
Using offline shadows provides a straightforward way of backing-up your databanks. Once you have completed the backup and set the shadow online, all operations performed on the master databank are applied to the shadow automatically.
Databank shadowing is entirely invisible to an application. This means that shadows can be added to existing applications. No special handling is needed to access tables in a shadowed databank.
A databank must support transaction control, see Transaction Control, to be shadowed.
Databanks with the WORK option cannot be shadowed, because shadowing requires transaction handling.
The databank to be shadowed cannot be used by any other users while a shadow is being created.
The shadow name cannot be the same as the name of the master databank, of any other shadow, or of any shadow that has been transformed to a master.
SYSDB and Shadowing
Because the SYSDB databank holds all the data dictionary information about your database, protecting it with shadowing and/or backups is essential.
Otherwise, if SYSDB is lost, the whole database will be unreachable.
SQLDB and Shadowing
Shadowing SQLDB is not allowed, as it is a TEMPORARY databank, and not needed because SQLDB only contains temporary data.
When you create a shadow for a databank, all tables and indexes in the databank are copied to the shadow.
Creating a shadow for a large databank may take some time, and thus should be carefully planned.
When you alter a shadow to a master, it only affects the Mimer SQL data dictionary. The databank file names are not changed. The databank cannot be used by any other user when a shadow is being transformed into the master (this is not very likely to happen since this function is normally used when the master has been lost or damaged).
We recommend that you take conventional backups as a supplement to databank shadows to protect data in the event of a crash that destroys both the master and the shadow.
Because operations are not interrupted when shadow-backups are taken, and because Mimer SQL databanks are automatically reorganized, you get true 24 hour-a-day operation.
When a shadow is dropped, the file where the shadow is stored is usually deleted from the file system. If it is not deleted, you can use operating system facilities to delete it.
The databank cannot be used by any other users while a shadow is being dropped.
Mimer Information Technology AB
Voice: +46 18 780 92 00
Fax: +46 18 780 92 40