Jun-ichiro itojun Hagino wrote:
When you have two connectivity that have different reachability, address selection and routing have to be appropriately controlled to avoid connection failure.but the above problem is not specific to VPN, right? - you have some connectivity from your laptop - either directly from your laptop or via some router(s) - some connectivity is restricted compared to another, in some way such as (a) bandwidth (b) price (c) NATed (d) slow (e) inefficient path due to tunnelling (f) limited reachability (g) you name it. it really is a policy routing problem. you have to solve it WITHOUT global knowledge. the way you (and probably other guys) are proposing with "distribution of policy table" is, using god's point-of-view.When you have two connectivity that have different characterisitic, such as bandwidth, price, delay and you name it, isn't it clear that people want to use their two connectivity as they wanted to. Of course they don't have global knowledge, at least they have knowledge of their own network.so you know that "local optimization != overall success". then why are you spending so much time trying to make incomplete local optimization?
They may want local optimization for their success. Though it may be incomplete, if it brings them a kind of success for their purpose, why other people can stop it ?