On 28.07.2016 22:40, Olle E. Johansson wrote: > Does anyone know what the problem they are trying to solve is? > (Still trying to find information)
Without taking any position of the usefulness on TCP proxies, and as an individual.....: They are deployed purely for performance reasons. It's a highly competitive area with multiple the vendors and the details of the methods is kept secret and covered by NDAs. But in general, the key belief is that the unique traffic characteristics in a mobile network makes a mobile network optimized congestion algorithm perform better. And also that the reduced latency between proxy and mobile helps. In addition, a proxy may also have congestion state feedback from the mobile network to further increase performance. The draft-johansson-cc-for-4g-5g may give some further insight into why a mobile network optimized congestion algorithm may perform better. And also that the issue is not too many retransmits but maybe to few due to bufferbloat.... Best regards Frode Kileng
