Generating globally unique mostly-sequential keys for DB rows across multiple processes
Note: This agorithm was co-authored by Phillipa Berresford. I wanted my domain classes to be able to reference each other by Id without having to add association properties - as I felt having all of this navigation in the model made it messy. I'd much rather have association properties within my aggregates (PurchaseOrder to PurchaseOrderLines) but not elsewhere. This was okay when I needed to add a reference to an object that already existed in the DB, but when I was adding a reference to a new object I needed to know the object's Id before Entity Framework Core saved it to the DB. As my tables were using DB assigned incrementing INT columns this was not possible, so it become obvious I was going to have to switch to GUID Ids. The problem with using GUIDs as clustered primary keys is that GUIDs are designed to be random. Whereas an auto-incrementing INT key will always result in new rows being appended to the end of a table, GUID keys would result in new rows being inserted...