I was thinking more about this, and our implementation reflects what Kent 
mentioned. The trader client does not send the user id directly to the business 
service because they live in two different security domains (Trader client in 
the bank, and the business service in the broker), and therefore the user id in 
both domains does not need to be the same. The Active STS running in the broker 
is who should map the "user id" claim sent by the trader client application 
(negotiated first with the passive STS) with the user id in the broker. In the 
current implementation, we have the same user id in both domains, but it could 
be different. 

Pablo.

-----Original Message-----
From: Nick Hauenstein [mailto:[email protected]] 
Sent: Thursday, December 10, 2009 8:07 PM
To: [email protected]
Subject: RE: StockTrader Sample Documentation Updates

I guess I could see that. In a way we do have multiple front ends, in the form 
of the different implementations of the stocktrader client. Once M2 is ready to 
rock, all front-ends should be able to connect to the same back end using their 
own passive STS (which in a way could represent multiple banks connecting to 
the same services).

The need for a passive STS in the client could still be justified, as the bank 
undoubtably has multiple web applications that would also require 
authentication at the same source.
 
I have versions of the swim lane in both configurations attached to the page in 
question [1]. I'll wait for a few more replies before I pull another swap.

- Nick Hauenstein

[1] 
http://cwiki.apache.org/confluence/pages/viewpageattachments.action?pageId=3474855

Reply via email to