ABAP Keyword Documentation → ABAP - Reference → Processing External Data → ABAP Database Accesses → Open SQL → Open SQL - Overview → Open SQL - SAP Buffering

SAP Buffer - Restrictions

Only transparent tables and pooled tables can be buffered. Buffering is not possible for cluster tables. There is no special restriction of the key length of the table when buffering, which means that keys in buffered tables can be up to 900 bytes long.

The following Open SQL statements bypass the SAP buffer implicitly and access the database tables directly, since the operation in question can only be executed on the database and not in the SAP buffer:

SELECT with the addition FOR UPDATE.

SELECT with the addition DISTINCT,

SELECT with aggregate expressions.

Open SQL statement with the addition CLIENT SPECIFIED when the client ID is not specified in a WHERE condition.

SELECT with JOIN expressions.

SELECT, if the WHERE condition compares a column with another column from a database table or database view specified after FROM.

Access to a table or view with individual record buffering without specifying (in the WHERE condition) all the equality conditions joined by AND for all the key fields of the primary key.

Access to a generically buffered area without complete specification of equality conditions joined by AND in a WHERE condition.

Open SQL statements with IS [NOT] NULL in the additions WHERE and HAVING.

Use of a subquery in a WHERE condition.

SELECT with FOR ALL ENTRIES in accesses to a table or view with generic buffering if the requirement to specify the generic area exactly (as specified above) is violated. The condition after FOR ALL ENTRIES cannot produce an OR relationship between multiple generic areas.

SELECT with the addition GROUP BY,

SELECT with the addition ORDER BY, if individual columns are specified as the sort key, and these columns are not a left-aligned subset of the primary key in the correct order.

After the invalidation of an entry in the buffer using a write statement, the next five reads that should have accessed the entry bypass the buffer of the current application server by default. The next read performed on the changed entry reloads it into the buffer and removes the invalidation. The number of reads that bypass the buffer before the reload is specified in the profile parameter zcsa/sync_reload_c.

Other versions: ~~7.31~~ | 7.40 | 7.54

Note

Bypassing the SAP buffer using the additions above should be noted for access to buffered database tables or views for performance reasons. To bypass the SAP buffer in the statement SELECT explicitly, the addition BYPASSING BUFFER should always be used. It is not enough to rely on the implicit behavior of the additions above.