RE: A new load balancing algorithm for discussion and consideration

["Marko Buuri" <marko@xxxxxxxxxx>]

>
>It's great to have new ideas. The -SH scheduler came this way.
>Do you have any interest in coding it up?
>

The changes I suggest to the existing WLC implementation are quite
simple logically and coding wise:
1) compare with (jobs+1)/weight, not jobs/weight. Now WLC would behave
like classic Shortest Expected Delay policy (SED).
2) return the first found server with current jobs=0, no matter the
weight. This implements the Never Queue policy (NQ).



Marko