@Allen Li I'm getting conflicting advice on whether to use

STRING

or

VARCHAR(N)

regarding query performance and memory usage. • You previously mentioned that I can always use

STRING

because "the length will be automatically determined by the actual data size." • StarRocks AI (Rocky), however, just told me the exact opposite: It claims that during queries (joins/shuffles), StarRocks pre-allocates memory based on the schema's max length, meaning

STRING

(treated as 65k+ bytes) causes massive memory overhead/OOMs compared to a strict

VARCHAR(N)

. It also said

STRING

hurts Primary Key index performance. My questions: 1. Does the execution engine pre-allocate RAM based on the schema's declared length or the actual data size? 2. Is there a real memory/performance penalty for using

STRING

instead of

VARCHAR(N)

? Thanks!