<div dir="ltr"><div class="gmail_extra"><br><div class="gmail_quote">On Tue, May 6, 2014 at 12:46 PM, <span dir="ltr"><<a href="mailto:ok@cs.otago.ac.nz" target="_blank">ok@cs.otago.ac.nz</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div id=":1dj" class="a3s" style="overflow:hidden">Observation 2: you probably *don't* need to know what is the absolute<br>
best room. You just need not to pick bad rooms. So you could have<br>
M managers each managing N rooms. When a person arrives, it picks a<br>
random manager and asks it what room to go to. If a manager has no<br>
suitable rooms, it sends a "try another manager" response. This<br>
reduces the amount of traffic each manager has to handle by a factor<br>
of M. This is very likely *better* than TM would gives you.</div></blockquote></div><br>Observation 3: if we can hash the picker in some way, we can do even better. Ask *some* of the managers randomly according to the hash of the picker. Choose the manager that gives the best answer. If a manager doesn't answer within a time frame, he is marked as overloaded and ignored. This provides fault tolerance in addition to the observations by Richard.<br>
<br clear="all"><div><br></div>-- <br>J.
</div></div>