Re: Disk cache

On Apr 20, 2006, at 00:26, Radu Hociung wrote:
> Arvid Norberg wrote:
> > I would say it's a huge difference. This is not too hard to calculate
> > analytically though, no real need to test it.
>
> Right. I gave it a shot in the other subthread with the
> quarter-by-quarter-random vs. piece-by-piece-random strategy  
> comparison.
> Is that what you had in mind?

No. Saying that it would be easy to show analytically was not very  
accurate though. Just to formulate how to measure performance is  
quite difficult. My best attempt would be something like:

the goal is to have as high probability as possible that the overlap  
between two randomly selected peers is as small as possible.

Striving for that would result in peers being able to exchange data  
as much as possible.

Your suggestion is basically just a way to increase the piece sizes,  
which will decrease the granularity of the current algorithm.


--
Arvid Norberg



-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642