Recursion vs Redirection<Pine.LNX.4.10.10110041911150.27989-100000@thread.crystalballinc.com>

[Oliver Spatscheck <spatsch@research.att.com>]

Eric,

	the only problem with a recursive approach is that it requires changes
to the existing infrastructure and adds additional overhead to the CDN
receiving the inital request (For example, if a flash crowed arrives
the CDN receiving the inital request has to track all the state of the
outstanding requests to other CDNs etc..). As pointed out before I
think whatever protocols we come up with should support both.

Oliver


Eric Dean writes:
 >  
 > A few months ago, I mentioned the concept of using a recursive DNS query
 > for CDI rather than CNAMEs.  CNAMEs create problems whereby a peered CDN
 > experiencing QoS problems can adversely affect content requests until a
 > TTL expiration occurs.  This can create apprehension when a CDN operator 
 > does not have strict control of a publisher's content.
 > 
 > However, CDI via recursive DNS queries that transpose the client's IP
 > address into the additional section preserves all CDN request routing
 > information while providing a different level of administrative control.
 > 
 >   
 > Eric Dean
 > President, Crystal Ball Inc.
 > W 703-322-8000
 > F 703-322-8010 
 > M 703-597-6921 
 > 
 > 
 >