I am looking to introduce an organisation wide unique document identifier across servers, multiple replicas, that is preferably sequential on a database per database/server per server basis but not essential.
My thoughts were to create some kind of server numbering document which controls this. Therefore we say that for server A the number range is 1000-10000, Server B 10001-19000, Server C 19001-n.
Hence when a document is created in a database it is numbered based on the server.
Anyone got any thoughts or any other ideas about cross replica unique numbering?