RE: [security-services] SAML Browser Profiles Metadata

From
Jahan Moreh <>
Date
2003-04-24T20:25:14+00:00
ID
Thread
RE: [security-services] SAML Browser Profiles Metadata
If you have not 
already reviewed draft 02, please see the new, improved draft 03 available 
at

http://www.oasis-open.org/apps/org/workgroup/security/download.php/1750/draft-sstc-saml-meta-data-03.pdf

Thanks,

Jahan

----------------
Jahan Moreh
Chief Security 
Architect
310.286.3070 

  
-----Original Message-----
From: Jahan Moreh 
  [mailto:]
Sent: Wednesday, April 23, 2003 6:02 
  PM
To: SAML
Subject: [security-services] SAML Browser 
  Profiles Metadata

  
Colleagues - 
  

  
As promised, I have 
  published draft 02 of SAML BrowserProfiles Metadata. This draft heavily 
  borrows from Liberty 1.2 metadata spec, draft 1.0-06. The documents are in 
  Kavi and available for review. Please see http://www.oasis-open.org/apps/org/workgroup/security/download.php/1734/draft-sstc-saml-meta-data-02.pdf (an MS word version is also available). I 
  have also uploaded the schema document that appears in section 3 of this 
  document as an xsd file (see http://www.oasis-open.org/apps/org/workgroup/security/download.php/1736/draft-sstc-schema-meta-data-02.xsd.xml)

  
 

  
Below I attempt to 
  answer some questions that may come up:

  

  
 

  
Where is Source ID for 
  Artifact source?

  
Per Liberty specifications, the source ID is a SHA-1 hash of the 
  provider ID, which is a required attribute of the 
  source

  
 

  
Where is the 
  designation for Issuer?

  
The issuer of a SAML assertion MUST have the same value of the 
  provider ID.

  
 

  
Why would a 
  destination (Service Provider) that supports both browser profiles have to 
  provide two descriptors?

  
This is required to avoid designating a new element 
  "ArtifactReceiverURL". I.e., we have overloaded AssertionConsumerURL for both 
  browser profiles.

  
 

  
Where is the 
  designation for NameIdenifierFormat?

  
It is not explicitly designated. It can be specified 
  in the catch-all "Extension" element.

  
 

  
What happened to the 
  various trust models?

  
In the interest of time I have not specified trust 
  models. Given the practical experience with two interops, it appears that 
  exchanging SSL certificates (both client and server) is the de facto trust 
  model.

  
 

  
 

  
Thanks,

  
Jahan

  
----------------
Jahan Moreh
Chief Security 
  Architect
310.286.3070