Hostname strategy in an elastic environment

I’d like to know how people using cloud environments such as GCE or AWS name their servers.

There’s plenty of discussion around on how this was done in non-elastic environments and I was always a fan of e.g. webserver01

However, these days in the world of elastic environments and auto-scaling I’d like to know if people still follow this methodology?

Do you go to the rather complicated process of booting a server, looking up what other servers of your type exist, work out your index and name yourself appropriately?

Or just stick with whatever your provider has assigned you with and use other methods to distinguish your servers e.g. tags

Is there somewhere in between?