Today, Amazon Redshift introduced additional performance enhancements that speed up string-based data processing by 5x to 63x compared to alternative compression encodings such as LZO or ZSTD. Amazon Redshift achieves this through vectorized scans over light weight CPU-efficient dictionary-encoded string columns that allows the database engine to operate directly over compressed data. These techniques are optimal on low cardinality string columns (CHAR or VARCHAR). Low cardinality string columns are columns that have up to a few hundred unique string values.
Source:: Amazon AWS