Performance implications of using CUID as primary key in MySQL

[emielvanliere]

I have a question regarding of using CUID’s in MySQL and their performance characteristics.

I can see in the Prisma source code that they store cuid's as VARCHAR(25) for id columns, which implies (at least) 25 bytes per id as opposed to 4 or 8 bytes for (big)int. And these will of course also be used for join-tables and foreign keys.

Although I think nowadays we shouldn’t be concerned that much about disk-space, maybe even index-size; what concerns me is the efficiency of index-lookups for VARCHAR. I understand CUID’s are monotonically increasing which is beneficial for b-tree indexes but I expect that an integer index is (much) more efficient than varchar.

I couldn't find any benchmarks regarding cuid (or similar monotonically increasing string id's). Does this scale to millions (or billions) of rows, or does it fall apart at some point? Thanks!

[pendenaor]

Maybe because js Number types is not reliable when dealing with integer greater than 53 bits, see https://www.ecma-international.org/ecma-262/6.0/#sec-number.max_safe_integer. Just my two cents.

[ericelliott]

It scales well to billions of rows. We used it in Adobe Creative Cloud.