[Tilesathome] Cutover to new server? When?

[Martijn van Oosterhout]

On Fri, Jul 11, 2008 at 1:01 PM, Florian Lohoff <flo at rfc822.org> wrote:
> Have you had a look at the complexity distribution?

Sure, it's logarithmic. An increase of one in complexity has 10 times
as many nodes, or something like that.

> So i have a hard time finding something above 10 and below 5.

If it's more than 10 it's Really Bad (the rotterdam tiles are like
that, for example). Think Richter Scale :)

> Breaking down an infinite number of objects into in finite scale
> will always have its problems.

Ofcourse, but you can approximate pretty good.

> Just the examples above make a difference between 100MByte needed and
> probably something like 4-10GByte needed.

Which correlates well with the factor-of-10 relation.

> A more finegrained and more deterministic resource requirement would be
> appreciated.

More finegrained? You think a scale 1-10 million would be more useful?
I put it on a logairthmic scale because that makes it easier to
configure. In the end it's the scale that matters, not the exact
number of nodes. It would be better to have the number of ways, but
that's much harder to calculate.

Have a nice day,
-- 
Martijn van Oosterhout <kleptog at gmail.com> http://svana.org/kleptog/