Alex Rousskov píše:
>
> On Mon, 13 Sep 1999, David Luyer wrote:
>
> >
> > Milos Prudek wrote:
> >
> > > Let's say I need priority access for a single IP (let's call it VIP).
> > > Whatever bandwidth that IP asks for, it will get it. If inactive, other
> > > IPs will get unlimited access. Seems to me I can't set delay pools to do
> > > exactly that.
> > >
> > > Given our capacity of 3000 B/s, I can say that VIP can have no more that
> > > 2500 B/s, and other IPs can have no more than 1500 B/s. That's fine when
> > > both VIP and IPs are surfing, but when VIP quits, the IPs are still
> > > limited to 1500 B/s, right?
> >
> > Quite correct. HTTP is stateless. There's no definition of when VIP is
> > surfing or not surfing to start with.
>
> I have to disagree in this particular context.
>
> IMO, the definition of "when VIP is surfing" is "when there is a pending
> request from VIP". Technically, it should be possible to limit others'
Well, it could be said like this: When VIP does NOT use his allocated
badwidth (for instance when VIP uses less than 60% of his allocated
bandwidth), let other IPs get his share.
-- Milos PrudekReceived on Mon Sep 13 1999 - 23:41:32 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:48:24 MST