problems with VO certificate authorities
Roy Williams
roy at cacr.caltech.edu
Fri Oct 27 10:56:44 PDT 2006
I believe there is a case to be made that the VO security structure should include both patterns:
the Certificate Store vs the Browser-plus-Certificate. This being that place that the "warrant" or
certificate is located: remote or local.
The Store mechanism has a well-secured remote machine that issues "proxies" on demand, which can be
used on behalf of a user. A Store can service a grid of machines that all need various trust
assertions. However, it needs to be continuously available. Also, it adds another component to the
service structure, a component that must communicate securely with others, and thus adds software
burden.
The Browser mechanism holds the certificate locally, i.e. a laptop. It relies on physical security
of the machine that holds it, although some browsers can ask for a password as well as physical
access. It can communicate securely with a server, and can handle the certificates in a
user-friendly way.
Perhaps the best argument is the upcoming AJAX and JSON-RPC applications that bring great power to
the browser itself.
Google Maps/Sky is/will be an AJAX application, I suspect that soon enough Google Sky will become a
prime viewing platform for the virtual sky. It would be nice to have certificate-enabled access to
sequestered data.
More information about the grid
mailing list