注:请转载后注明出处
【一些理由和启示】
The number one goal of viability today is clearly a major driver for this protocol
development. With the understanding that middleboxes and firewalls would typically block
or dramatically degrade any transport based on formats other than TCP or UDP, we will not
even consider revolutionary protocols.
【目标1中的对当前的适应能力是这个协议发展的主要驱动,考虑到哪些中间设备和防火前典型的阻塞或者戏剧性的降级任何基于其他类型的传输,而不是TCP、UDP,这里将不考虑一种新的变革性协议】
All parties would have preferred to satisfy goal 2 (re: congestion based packet loss on one
stream, impacting several streams) by evolving TCP to avoid headofline blocking, but we
found there was no apparent way to circumvent TCP’s inorder delivery interface.
Modifications are plausible, but could not be widely deployed until kernel changes were in
place, and that was viewed as a showstopper. In addition, more significant modifications
of TCP would potentially be blocked by today’s middleboxes, which again would take many
years to evolve.
【所有参与者都优先目标2通过金华TCP来避免线头阻塞,但是我们发现目前并没有一种明显可行的方法来战胜在接口上TCP的顺序的传输,修改是貌似可行,但是如果内核不准备就绪这个肯定也无法广泛地传播,这是一个很突出的问题。额外地,针对TCP较多的重要的改动将被当前的中间设备所阻挡,而这些设备没有个几年也不会被升级。】