Show TOC

Background documentationConverter Locate this document in the navigation structure

 

The converter assigns logical data pages to physical storage locations (converter block address). The database system accesses data pages in the data volumes using their converter block address. The converter block address consists of the volume ID and the offset in the data volume.

Example Example

Converter entry

271397 P 1/0000271404

The logical page number 271397 is a permanent (P) data page. It is stored in the first data volume at position 271404.

End of the example.
Data Page Types

Type

Description

P

Permanent data page

S

Data page in an SAP liveCache database

T

Temporary data page

The database system imports the complete converter into the converter cache when the database starts.

If the database system fails to find a page number in the data cache, it searches for the page number in the converter cache and uses this to calculate the physical position of the data page in the data volumes.

At each savepoint, the database system writes the contents of the converter cache to the data area and gives the stored converter a version number. The converter saved at the next savepoint is then given the next highest version number. Each converter version corresponds to a consistent state of the database.

The converter uses the shadow storage concept. The database system stores the database state most recently written into the data area (most recent converter version) until the next savepoint. If the database user wants to change data pages, the database system first imports the data pages into the working memory and makes the changes there. The database system then writes the changed data pages to a new location in the data volumes at the next savepoint. Only at the end of the savepoint does the database system release the old data pages in the data volumes for overwriting.

The same logical page can be stored at different physical locations at different times.

Snapshot of the Data Area

To create and manage snapshots, the database system uses a shadow memory concept: a snapshot saves the current converter. When you create a snapshot, the database system saves a link to the current starting point in the converter in the restart information of the database and flags all the physical pages in the data area whose addresses are in the converter. The database system cannot overwrite these pages as long as the snapshot exists.

Data changes that you make after the snapshot are written by the database system to other physical pages in the data area. If you delete a table, the database system does not physically delete this table, but only flags it as deleted (you can no longer access this table with SQL statements). In this way, a new version of the converter is gradually created, which with time differs increasingly from the version of the converter that was saved in the snapshot.

If you reject all changes you have made since the snapshot, then the database system deletes the starting point in the new version of the converter and replaces it with the starting point that is saved in the snapshot. The database system releases all pages that were changed after the snapshot for overwriting again.

If you delete the snapshot, the database system releases the flagged pages in the data area for overwriting.