Re: PolyMix-3 rule: Fill phase request rate vs Peak rate

From: Alex Rousskov (rousskov@measurement-factory.com)
Date: Thu Jun 07 2001 - 13:16:47 MDT


On Thu, 7 Jun 2001, K Ho wrote:

> The Third Cache-Off Official Report (Section 4.1)
> specifies that "...the request rate during the Fill
> phase must not be greater than the peak rate (as used
> in top1 and top2)."
>
> What is the reasoning behind this rule? Is there some
> advantage in using a high request rate for the fill
> phase?

Good question! Yes, there may be an advantage for some products. The
reasoning behind fill rate limit is now documented at
  http://www.web-polygraph.org/docs/workloads/polymix-3/#Sect:1.2

    ``PolyMix-3 limits fill request rate to peak request rate to prevent test
    participants from specifying very high fill rate that causes the cache
    to reject or bypass some of the incoming fill traffic, effectively
    reducing the amount of content stored by the cache. Some products
    used this (now illegal) trick to cache less data during the fill and
    optimize their dataplacement layout for the measurement phases.
    Ideally, Polygraph should check how much data is actually cached
    instead of imposing artificial request rate limits.''

Thanks,

Alex.



This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:19 MDT