I think the only reason we're using Crowd now is to map LDAP groups into Jira.
We've had Crowd deployed for two years to handle web app auth. Configuration, deployment were a bit painful.
We're pulling it out and writing our own - seems to be the only way to get necessary features such as delegated administration which have been on the Crowd roadmap for years now.
JIRA 4.3 and Confluence 3.5 will make it much easier to connect to Crowd. No editing of files required.
We don't have write support for the Posix schema scheduled for Crowd, I'm afraid, so no promises on that front.