Scaling Issues

There are certain points of concern with regards to how the system is currently setup where we will have to watch out for performance and scaling reasons.

Credentials Storage

Currently the entire authentication mechanism uses htpasswd based files for storing username and passwords. The initial design chose this path because it was the easiest to integrate with Subversion/Apache access. So anything that changes here has to take into account authentication not only for Trac but for Subversion/Apache as well.

• The apache/svn access currently lives outside the WSGI stack of ClueMapper which makes it the limiting factor in choosing a new credentials storage mechanism (ie we can pretty much make ClueMapper deal with whatever we want)
• Several hopeful options for integration would be...
  • SQL based credentials lookup (seems most practical for apache/svn but will grossly complicate deployment)
  • OpenID
  • Google Accounts