Sharing performance testing findings

[Cantor, Scott]

On 4/5/19, 9:26 AM, "users on behalf of Jason B. Rappaport" <users-bounces at shibboleth.net on behalf of jasonrap at princeton.edu> wrote:

> 1. We changed the order within our Shibboleth IDPs (attribute-resolver.xml) so the first DC they would hit for attribute
> resolution would be different 

Seems like some more memory combined with connector caching would probably eliminate a ton of the load, though it's not a real world scenario regardless. Load testing an IdP gives you an incredibly distorted view of the actual load an IdP tends to be under, unless you're using it for webmail as a basic-auth proxy.

-- Scott