The problem is the unique IDs are GUIDs generated independently (and offline) by the client and then passed to the database; we need to make sure that no event is processed twice.
Currently there's one big log table and a uniqueness constraint on the GUID field. I'd love to partition the table out but until I can identify a clean route to satisfying that requirement I'm keeping one large log.
Though, now that I think about it, maybe partitioning on the first character of the GUID string is feasible, as we can ensure uniqueness per-table then
Currently there's one big log table and a uniqueness constraint on the GUID field. I'd love to partition the table out but until I can identify a clean route to satisfying that requirement I'm keeping one large log.
Though, now that I think about it, maybe partitioning on the first character of the GUID string is feasible, as we can ensure uniqueness per-table then