Re: Polymix-3 vs Polymix-4

On Thu, 11 Mar 2004, Bryn Reeves wrote:

> I've been asked to write a brief report on the differences between
> polymix-3 and polymix-4. The only addition on the feature lists is
> that of DNS support and DNS addresses within experiment URLs.
>
> Are there other areas in which polymix-4 is significantly better in
> terms of realism, or accuracy of the content/request models it's
> using?

For a more complete list of changes, please see:
http://www.measurement-factory.com/results/public/cacheoff/N04/report.by-meas.html#Sec:Workload

DNS addresses and better byte hit ratio simulation are probably the
most important improvements when it comes to realism. Byte hit ratio
improvements and POST/HEAD requests also affect the average response
size, which may have significant effect on proxy performance. Aborted
transactions punish proxies that do not deal with such transactions
well.

If you are developing a new workload, start with PolyMix-4 and
consider adding recent features like SSL traffic and, if your proxy
looks at content, content/filtering-specific features. In other words,
tailor PolyMix-4 to your environment.

Alex.
Received on Thu Mar 11 2004 - 12:44:50 MST