Alternative ID type?
I've read that auto-incrementing IDs can present scaling problems when a project is big enough to begin partitioning tables. Massive offers UUID as an alternative, which accommodates IDs to the partition problem, but UUIDs can present their own performance issues since they aren't sequential, and according to what I've read, in the general case there are benefits to having at least pseudo-sequential primary keys.
There's a library called
cuid that presents some of these arguments, and a solution:
Would you consider making Massive configurable to allow the
saveDoc API to use
cuid, or to make the ID column be of type
text and let the user insert their own ID? I'd be interested in hearing why or why not.