The Data Base Handler's (DBH) buffer and container technology represents an important function for increasing throughput. Buffers and containers are main-memory areas that are used to store data temporarily.
The SESAM/SQL DBH's most important buffers and containers are
User data buffer - for user data
System data buffer - for system access data
log buffers, used for transaction logging and logical data backups
the cursor buffer, used for the intermediate results or retrieval statements
the plan buffer, used for SQL access plans
the transfer container, used for SQL scans and inquiry and response areas of logical files
the work container, used for internal statement formats
The aim behind the buffer and container technology is to minimize hard-disk I/O: write jobs that would normally be carried out block by block can be collected and written to the disk in one operation.
Since main memory is not available in unlimited quantities, not all buffered data blocks can be kept resident in memory. The data blocks in the buffer areas and the work container are therefore displaced when a buffer overflows. The DBH's buffer administration feature decides which blocks are to be displaced; the decision is based on criteria intended to achieve the best possible throughput. For example, the DBH sets up separate buffer areas for data with differing displacement behavior, but also differentiates when displacing blocks within the individual buffer areas. In doing so, it avoids the displacement of blocks containing frequently required information in favor of blocks containing information required less frequently.
The sections that follow describe which data is stored temporarily in the individual buffers and containers.