Hacker News new | ask | show | jobs
by scott113341 2608 days ago
Not strictly related to this article or your question, but generating ordered UUIDs can sometimes bring some performance improvements (at least in Postgres) on index reads/writes [1].

[1] https://www.2ndquadrant.com/en/blog/sequential-uuid-generato...

1 comments

Similarly for SQL Server sequential UUIDs can reduce fragmentation, which can save some IO, storage space, and memory load, particularly if the UUID used a table's clustering key (most cluster by their PKs by default).

Unfortunately without a bit of a hack NEWSEQUENTIALID() can only be used in TSQL as a default constraint, which can be an irritation.

And you are often better off using something else than a UUID for a table's clustering key, to help common range queries and reduce clustering key size, even if it isn't the PK or otherwise particularly unique.