> The latter question is a real concern, but we won't know until we go > and collect some data. When we get measurements for these sorts of > things, it's usually from services that have the resources to acquire > the measurements. At the same time, those services likely have the > resources to have a CDN and so probably will have less need for > caches.
It seems the right people to work with on this are the people at rural/edge/mobile ISPs, as these are the people who will know the size of their caches, the hit rates, the amount of bandwidth they take off their transit links and the amount of money they'd have to raise from their customers to pay for increased peering/transit costs if their caching proxies all broke (extrapolated out over the expected transition time, of course). Providers of web services themselves don't seem in a good position to calculate the value of caching http proxies, as they aren't the ones benefiting except in an indirect latency based way. _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
