The term "useful" here does not mean any economic price but technically. I think those cases that someone sells the information to another for high price is out of our discussions. This is just a cheating problem and it may be prevented by some mechanisms(maybe we can take into account). The term "useful" meaning for client is that it can help client for improving performance or traffic optimization.However clients may lie and ask more than what they really need and it can be hard to tell if they are lying(maybe client application-classification is good choice, one kind of application needs some fixed or less amount of "useful" information for it, some like access-control mechanism). This all can be considered in ALTO protocol.
2009-12-31 熊淼 Xiong Miao Mobile Life and New Media Lab Beijing University of Posts and Telecommunications Addr:No.10 XiTuCheng Road Haidian District,Beijing 100876, P.R. China 发件人： Richard Alimi 发送时间： 2009-12-22 22:58:54 收件人： Xiong Miao 抄送： Enrico Marocco; [email protected]; standardMINE 主题： Re: [alto] New draft notification:draft-wang-alto-privacy-load-analysis-00 On Tuesday 22 December 2009 4:59:56 am Xiong Miao wrote: > Thank you for your categorization, for it make me more clear about the > "privacy issues". > But i'm still puzzled about how to define information that SHOULD NOT > provide for ALTO servers since the information may be useful for one > client but not useful for another. It's like an access-control issues. > ALTO protocol needs something to do with this. I agree that the ALTO Protocol should allow for access control *for requests to the ALTO Server*. That is, it should handle (3a) and (3b) in the categorization (which also entails encryption to prevent snooping on the wire). This is doable with existing techniques such as HTTP Basic/Digest authentication, and SSL/TLS presuming that we stick with a protocol built on top of HTTP. Can you clarify what you mean by "useful"? The term "useful" is a very general term that could encompass many uses, both legitimate and otherwise. For example, an ALTO Client may have a business agreement to obtain detailed network information from a network provider via ALTO, and the agreement may state that the ALTO Client cannot share the information with a 3rd party. The ALTO Client may consider it "useful" to ignore the legal agreement and sell the information to a 3rd party anyways for a large sum of money. This is the primary reason there is a distinction between (3a-b) and (3c). It is to allow the information to be protected in transit, and to protect it from being *directly* distributed to an unauthorized party. However, enforcing access control once an authorized ALTO Client has received it would entail other mechanisms (e.g., DRM) as Enrico has mentioned. Another way to summarize the difference, is that the ALTO Protocol provides access control for requests to an ALTO Server, but it *does not* provide access control for the ALTO information itself. -- Richard Alimi Department of Computer Science Yale University
_______________________________________________ alto mailing list [email protected] https://www.ietf.org/mailman/listinfo/alto
