Provide a way to fetch asyncrounsly large Metadata and manage its content with a Cache system

[peppelinux]

This is not an Issue but my thoughts about some improvements that could sounds more efficient in production environment. I wrote this post with my latest ideas, then I should read some other posts over the net to integrate them into this.

Expected Behavior

Code and Documention about a metadata class like saml2.mdstore.MetaDataExtern, for example, to fetch metadatas from a cache system like nosql (memcached, redis...) or filesystem.

I think that download should be scheduled when dealing with large data set, as federation's metadatas. Thus I think that this task could be also be done in a integrated way with the same Class, specialized methods.

Current Behavior

At the moment in pysaml2 metadata are donwloaded and stored in a specific filesystem path or downloaded at every http request. I put some notes here: https://github.com/peppelinux/Django-Identity/blob/master/djangosaml2_sp/djangosaml2_sp/djangosaml2_sp/sp_pysaml2_shibidp.py#L143 (nothing important, it want just to let newcomers to have some reference).

Possible Solution

Just start the discussions, elements will be collected in future.

Additional resources

https://github.com/knaperek/djangosaml2/issues/113#issuecomment-401441848 https://github.com/IdentityPython/pysaml2/blob/master/src/saml2/mdstore.py

[peppelinux]

I'm looking for a better solution using pyFF, with the possibility to get information on demand from it.

""" The pyffd tool starts a metadata server with an HTTP-based interface for viewing and downloading metadata. The HTTP interface can produce XML, HTML and JSON output (aswell as other formats with a bit of configuration) and implements the MDX specification for online SAML metadata query. """

see https://github.com/IdentityPython/pyFF

[peppelinux]

useless once MDQ and pyFF have been implemented.