@Rocky is there much difference between STRING and something like varchar(1048576)?

In StarRocks, the primary difference between

STRING

and a large

VARCHAR

like

VARCHAR(1048576)

is the maximum allowed length and how the system manages memory during query execution. 1. Maximum Length *

STRING

: In StarRocks,

STRING

is essentially an alias for

VARCHAR(65533)

. Its maximum length is 65,533 bytes. *

VARCHAR(1048576)

: Since StarRocks v2.1, the maximum length for

VARCHAR

has been increased to 1,048,576 bytes (approximately 1 MB). * Note: Both types measure length in bytes, not characters. In UTF-8, a single character (like a Chinese character) can take 3–4 bytes. 2. Storage vs. Memory Performance * Storage: There is no difference in disk storage. Both are variable-length types, meaning if you store the word "Rocky" in a

VARCHAR(1048576)

column, it only consumes the bytes needed for that string, not the full 1 MB. * Memory Allocation: This is where the difference matters. In some query scenarios, StarRocks pre-allocates memory based on the defined length of the column rather than the actual data size. * Using

VARCHAR(1048576)

for every column can lead to higher memory overhead and potentially "out of memory" (OOM) errors during complex joins or aggregations because the execution engine may prepare for the worst-case scenario (1 MB per row). * Using

STRING

(which is 64KB) is safer for memory than the 1MB max, but still larger than a targeted

VARCHAR(N)

. 3. Summary Comparison Feature |

STRING

|

VARCHAR(1048576)

Max Size | 65,533 bytes (~64 KB) | 1,048,576 bytes (~1 MB) Storage | Actual data length | Actual data length Memory Risk | Moderate | High (if used excessively) Usage | General-purpose text | Very long logs, JSON, or large text blobs Recommendation * Be precise: Use the smallest

VARCHAR(N)

that fits your data (e.g.,

VARCHAR(255)

for names). * Use

STRING

: If you know your data is under 64KB but don't want to define a specific limit. * Use

VARCHAR(1048576)

: Only when you specifically expect data to exceed 64KB. If you need even more space, you should consider the

JSON

type or storing data in an external system. References • knowledge/usage/best-practice.md • sql-reference/data-types/string-type/VARCHAR.md • faq/table_design_faq.md • faq/Others.md • sql-reference/data-types/string-type/STRING.md • sql-reference/data-types/string-type/BINARY.md • faq/Sql_faq.md

@GregM please be aware of https://github.com/StarRocks/starrocks/issues/73374 and

transform_type_prefer_string_for_varchar