tablescan

TableScan is a database query execution operation that reads every row from a table, typically in a sequential manner, without using an index to locate specific rows. It represents the most straightforward form of data retrieval and is often contrasted with index-based access methods, which target a subset of rows via a lookup structure.

Table scans are commonly chosen when the table is small, when there is no usable index for the given predicates, or when a large portion of the table is required by the query (for example, selecting many or all columns). In many systems, predicates may be applied during or after the scan; predicate pushdown can allow the storage layer to filter rows early and reduce I/O, while projection can limit the number of columns read, especially with columnar storage.

Performance considerations center on data volume and I/O bandwidth. The cost of a table scan generally grows with the amount of data read and can be I/O-bound on disk-based storage. In memory or cached environments, CPU and memory bandwidth become significant factors. Many engines parallelize scans by partitioning a table and distributing work across multiple workers, which improves throughput on large tables. Data layout and storage format influence effectiveness: row-oriented formats execute straightforward scans, while columnar formats can accelerate scans by reading only the needed columns. TableScan remains a fundamental operator in query execution plans, used when no selective index is available or when most of the table must be accessed.