Slide 4 - Proposed Solution
Not
being one to make dire predictions without offering a "solution", I offered
this slide. It represents what many folks (OK, not really MANY, but
enough of the folks at SIGCOMM anyway) will recognize as a "time sequence
graph" showing TCP's slow start behavior. I said that, because of
my research into transport protocols, I believed that the beautiful theory
and practice from TCP could solve our problem for us. If industry
wants one student, then they must ACK that student somehow, which allows
that company to hire 2 students the next year. Once those 2 students
are "ACKed", they can hire 4, and etc. in the familiar pattern. This
would allow resources to be clocked in and out of the system in a deterministic
way. Of course, that leaves the question of the form that the ACKs
should take, which leads to...
Next Slide
This
page created with Netscape Navigator Gold
Not
being one to make dire predictions without offering a "solution", I offered
this slide. It represents what many folks (OK, not really MANY, but
enough of the folks at SIGCOMM anyway) will recognize as a "time sequence
graph" showing TCP's slow start behavior. I said that, because of
my research into transport protocols, I believed that the beautiful theory
and practice from TCP could solve our problem for us. If industry
wants one student, then they must ACK that student somehow, which allows
that company to hire 2 students the next year. Once those 2 students
are "ACKed", they can hire 4, and etc. in the familiar pattern. This
would allow resources to be clocked in and out of the system in a deterministic
way. Of course, that leaves the question of the form that the ACKs
should take, which leads to...