Prior to joining Atlassian, I used Crowd for authenticating customer users to a product that operated outside of our corporate network (in another data center). What I ended up with may meet your needs.
(1) I had an Active Directory that represented our internal user base. This consisted of two (2) Active Directory domain controllers used by most every service internal to our company and one (1) READ-ONLY active directory domain controller setup for the purpose of this scenario.
(2) I had another LDAP directory that represented all customer-users because I needed to be sure customers and internal users were never co-mingled–a management requirement.
(3) I setup Atlassian Crowd as the bridge between the two LDAP directories and then exposed the Crowd end points to the remote data center (which I then white-listed by IP address). A better solution would have been to use a VPN tunnel, but at the time that was not an option due to what we will just call "vendor relationship issues."
(4) Here's the important part. Because I pinned Crowd to authenticate first against the customer LDAP then against the read-only domain controller (with no ability to resolve to the other DCs), I was pretty certain that (a) customer accounts would be resolved first (and fastest), (b) no one could update information in the Active Directory via Crowd and (c) given my Active Directory group policies that locked accounts after three (3) attempts, I was pretty certain that my corporate VPN (which was also on Active Directory) was more vulnerable than my crowd installation when it came to external attacks.
(Disclaimer: What I just described was done before I joined Atlassian and I'm not entirely sure if it complies with Atlassian best practices...I just know it worked.)
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.