Hi,
1. I believe there is a maximum limit of 10000 users beyond which crowd's performance degrades and caching is not suggested.
Is there a limit for groups as well?
Currently, we are creating 2 groups per project because we are also managing subversion repository permissions with Crowd. We would have around 5000 projects which would mean 10000 groups, though users may be within the 10000 limit. Are there any limitations in crowd to handle this number of groups?
2. Also, is there a limit to the number of application instances of JIRA, Confluence and fisheye that 1 crowd instance can support to provide Single Sign On (SSO)?
3. When we connect a JIRA/Confluence/Fisheye application to 1/more crowd directories, does the application synchronize *all* the users and group memberships and groups of those directories and store it in its embedded crowd tables? Or is it only the users and the users' group memberships that are currently accessing the application?
Does this pose a limit to how many users and groups we can have in crowd per directory or per application that we are mapping?
4. If there any limitations for above questions, please specify what limitations are only related to slight loss of performance and which ones are completely unsupported by Atlassian support.
Understanding of these limitations up front is very important as we are planning out our enterprise roll-out and are taking major architectural decisions.
We are currently using Crowd 2.4.2 but we can upgrade if there are any enhancements related to this in the latest version of crowd, so please advise accordingly.
Thanks and Regards,
Archanaa