Re: Request for stats addition

Cassaela (
Thu, 5 Jun 1997 19:08:40 -0700 (PDT)

On Thu, 5 Jun 1997, Andy Church wrote:

> >On the issue of speed, I'm not sure how meaningful that is. If I
> >throw 30 machines at deschall for 10 hours a day, is the speed
> >based on the #keys checked in 10 hours, or the same number averaged
> >over 24 hours? The example claims a figure of "keys per second",
> >and that seems somewhat meaningless in this context.
> If the clients report back how long it took them to finish the
> keyblock they were on (I expect they must, else how does the keyserver
> know what size blocks to assign next time around?), then a valid
> keys-per-second figure can be derived, no matter what your deschall
> connectivity pattern looks like.

This was recently a rather large debate over at the SolNET effort. What
they had, which nobody seemed to like much, was an average key rate for
each domain. So, if you've got 20 machines crunching 700k keys/sec, the
stats would show your speed as 700k keys/sec, not 14000k keys/sec. Now,
the 700k keys/sec figure may be the more "correct" average, but I think we
can admit that the only reason we'd like to see speed and percentage
contribution is so we can scratch our geek egos and show off our platform
of choice..... in which case, bigger is always better. :)

Just thought I'd point this out so that, if this does get implimented, it
gets implimented "right" the first time and we can avoid the confusion of
changing it down the road.

Of course, now that I go back and reread the question, it seems like it
had very little to do with what I was just talking about. Speed and #keys
checked are two very different things; averaging your keys checked per day
over 24 hours is *not* your speed.


Duct tape is like the Force. It has a light side, a dark side, and it holds
the universe together.