The last two entries provided an overview of the peer DP role in SCCM 2007 and also discussed admin provisioning of a peer DP.  For this entry we will discuss 'on demand' peer DP provisioning.  It should be noted again that SCCM 2007 is still a beta product so these details remain subject to change.

'On demand' provisioning happens when a client is targeted with a deployment and none of the accessible DP's available to the client have the content available.  In such a situation, if there is a peer DP available to the client, that peer DP will be automatically added as a distribution point for the package.  In this way, the next time the client attempts to execute the deployment it should be available on the accessible peer DP.

Lets go a bit deeper into the process.  There are a few conditiona that must be true for 'on demand' provisioning to occur.

1.  There are no standard DP's the client can access that have the content available
         Note, this does NOT mean that no standard DP's have the content - it is required
         that at least one standard BIT's enabled DP is available with the content at the 
         site where the peer DP is located.  If this is not the case, the on demand request 
         will be delayed until the condition is met.

2.  There is at least one peer DP is protected and includes the clients boundary.

3.  The advertisement property controlling client fallback to non-protected distribution points
is enabled.  This property in described as follows in the advertisement GUI:  

         "If the client is in the boundaries of one or more protected distribution points, 
         allow the client to access the content only from the protected distribution points"

4.  The package must be configured to allow 'on demand' provisioning.

With all of this enabled the flow of events would be as follows....

-A package/advertisement are created.  The package is staged on at least one BITS enbled standard DP - protected or not.

-The advertisement is received by a targeted client.  The client attempts to find the content so it can run the deployment and makes a location request to the management point.

-The management point queries the database to determine a list of available distribution points that are available with the content and that the client can access.  Because the fallback flag is enabled on the advertisement, no distribution points are returned but the management point does determine that a peer DP is available in the protected boundaries of the client but just doesn't have the content.  Based on the the management point triggers addition of the peer DP to the distribution points list for the package.

-The client will receive a blank list of distribution points from this initial content request and the first attempt to run the distribution will fail.

-Behind the scenes the DP is added and policy is adjusted so the peer DP knows it has a package to download.  On the next policy evaluation cycle the peer DP learns of this policy and the software gets downloaded to the peer DP.

-The client tries again and again sends a contentlocation request to the management point.  This time the peer DP is returned to the client as an available source of the package and package execution continues.