Re: PolyMix-2 working set

From: Alex Rousskov (rousskov@ircache.net)
Date: Sun Dec 05 1999 - 19:14:26 MST


On Sun, 5 Dec 1999, Andy Ho wrote:

> I have a question about the working set of the PolyMix-2 workload.
>
> Let's say a cache obtains object X from a polyServer at time T1.
> Does that mean that X will not be accessed by any polyClient
> after T1+4-hr?

Good question.

Not exactly. Working set size is frozen after first 4 hours of an
experiment, but it is measured (internally) in the number of objects.
That is, the number of objects accessible at 4 hour mark becomes the
working set size. Thus, the "age" of the oldest object in WSS depends on
request rate. The latter is not constant during a PolyMix-2 experiment.

If T1 is the beginning of the "inc2" phase, than your statement is true
(within reasonable precision). For other T1s, your statement may not
hold.

Alex.

P.S. Freezing WSS using time makes the workload fair. Using
     number of objects internally, makes 4hour limit simulation
     feasible.



This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:10 MDT