{"id":"data-scan-skips-value-bytes","text":"`_scan_data_file` in the hash-index Bitcask reads each record's header and key but seeks past value bytes; it is still O(data_size) because headers must be parsed sequentially to locate record boundaries","truth_value":"IN","source":"entries/2026/05/29/topic-bitcask-startup-cost.md","source_url":"","source_hash":"","justifications":[],"dependents":[],"metadata":{},"explanation":{"steps":[{"node":"data-scan-skips-value-bytes","truth_value":"IN","reason":"premise"}]}}