OSM's future. [Openstreetmap]

[Lars Aronsson]

SteveC wrote:
> > Do you have stats about amount of resources used by people in "edit" 
> > mode and in "view" mode ?
> 
> per-session stats arn't recorded but as a rough guide tile saw 52,000
> hits yesterday from 320 different ip addresses.

What kind of information can you get out of the tile Squid?  Does 
it report the cache/hit ratio, and the min/avg/max processing time 
for the cached and non-cached requests?  My impression is that 
cached tile requests are served really fast, and I guess these 
numbers would hint at extending the cache expire time from 48 
hours to 14 days or more.  Can we think of a way to invalidate 
(the right part of) the tile cache when a map is edited?

Re your earlier posting, I agree that OSM can need those 7-10 
servers within a year or so.  Right now 2-3 servers (new ones) of 
OSM's own (not borrowed) could be more reasonable.  If people are 
going to send money, they need some kind of organization or 
foundation to send it to.

But this organizational/hosting/funding question should be kept 
separate from current runtime performance problems, since the 
latter has much more to do with the software.  If we use slow XML 
libraries instead of fast ones, there will be no end to how much 
hardware we are going to need.

Right now OSM is a bit slow and this keeps willing contributors 
away.  I want to recruit more people, but I really cannot show 
them this slow server.  As soon as the system gets faster, we will 
get more users that will consume the speedup.  In order not to 
become depressed by the constantly slow server, we should keep 
stats on how many users and map views we serve, as these numbers 
should steadily increase.


-- 
  Lars Aronsson (lars at aronsson.se)
  Aronsson Datateknik - http://aronsson.se