On Fri, 7 Jul 2000 Craig_Lowery@Dell.com wrote:
> When you say that the cache fill rate must be between 10% and 100% of peak
> rate, you mean the total rate experienced at the proxy,
Correct.
> not as written in the workload file, right?
As far as I can see, the peak rate specified in the workload file will
be the same as the peak rate experienced at the proxy. In other words,
benchPolyMix3.peak_req_rate specifies the peak rate at the proxy.
> Documentation in "preparation tips" is unclear,
> but it follows that this is the case. However, to be on the safe side, here
> is an example. Please validate that it is correct.
>
> 5 pairs, 2000 peak, 50% fill rate. That means a fill rate of 1000/sec. The
> fill rate in the workload file will be 200/sec, since each pair will
> contribute 1/5 of the overall fill rate. Right?
I am afraid you have to specify 1000/sec rate in the workload file.
We will try to clarify the documentation (suggestions welcome!).
Perhaps tracking the FillRate variable usage in polymix-3.pg will help.
As you can see, FillRate is used in relation to PeakRate only. The
latter is total request rate experienced at the proxy:
rate FillRate = 1000/sec;
...
float FillFactor = FillRate/PeakRate;
IIRC, the only *per-pair* parameter is ProxyCacheSize, and even that
will be converted to "total size" as PGL becomes smarter.
Alex.
This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:14 MDT