A QUIC look at HTTP/3

Posted Mar 14, 2020 6:19 UTC (Sat) by djc (subscriber, #56880)
In reply to: A QUIC look at HTTP/3 by Cyberax
Parent article: A QUIC look at HTTP/3

The protocol has been carefully designed to prevent ossification (including several greasing features), so it seems likely that they've bought themselves some time in that regard.

A QUIC look at HTTP/3

Posted Mar 14, 2020 6:23 UTC (Sat) by Cyberax (✭ supporter ✭, #52523) [Link] (2 responses)

QUIC is carefully designed to be middlebox-proof. But it won't prevent ossification of endpoints. And multihoming is kind of a feature that requires a lot of serious changes, so it's likely to go unimplemented in many stacks.

A QUIC look at HTTP/3

Posted Mar 14, 2020 11:36 UTC (Sat) by draco (subscriber, #1792) [Link] (1 responses)

Even if the specification says how to do multi-homing, that's no guarantee that implementations will support it properly since you can't force an endpoint to have multiple addresses.

A QUIC look at HTTP/3

Posted Mar 14, 2020 17:41 UTC (Sat) by Cyberax (✭ supporter ✭, #52523) [Link]

It's fine to have a single-address server endpoint, but they need to be able to talk to multi-address clients. I guess it would be the most used configuration.