[rt-users] How do you use Queues?

[Bret Martin]

I'm curious to know how others use Queues.  It was a bit hard to do a
comprehensive search on this somewhat vague subject in the list
archives, but from what I gather from the searches I did, there seem to
be three basic models that are widely used:

1. one queue per entity receiving service (think "customer")

2. one queue per entity providing service (think groups like
   like "IT", "engineering", "sales", "marketing")

3. one queue per general type of work (think "desktop support",
   "web design", etc.)

Have I missed any other ways of thinking about it?  Has anyone used one
of these, realized it was wrong, and switched?  Anyone have thoughts on
what makes it appropriate to use one model versus another?

Obviously #2 and #3 are similar since often groups are formed around
work types like "desktop support", "network operations", and so forth 
-- but they sort of come at queue naming from a different direction.

--Bret