|
On Tue, 19 Sep 2000, Thomas Proell wrote:
> Hi!
>
> > with LVS you don't have that choice. Each real-server is loaded
> > progressivley attempting to keep the load the same on all machines.
> > You can't cut off one machine at a certain load.
>
> So, first you have to define what you mean with "load".
ahem, yes. here load = number of connections
(this is all the director knows about).
> It's easy to change any scheduler that way that he doesn't
> route to a server with more than "n" connections, which
> is a (not very good) approximation of "load".
>
> I'm writing on a scheduler that should lower the traffic
> to a "very loaded" server, which is not easy since
> "load" is hard to define.
there was a thread here a while ago, where the design of load informed
scheduling was discussed. An agent on the real-server monitors load
(has modules for whatever you want to monitor) and sends info back to
a demon on the director, which controls ipvs with ipvsadm.
Unfortunately producing working code is hard and nothing has
appeared yet.
Joe
--
Joseph Mack mack@xxxxxxxxxxx
|
