[OSM-talk] OSM as Map Publishers (was Re: TomTom acqures TeleAltas)
osm.list at randomjunk.co.uk
Mon Jul 23 17:00:20 BST 2007
On 23/07/07, 80n <80n80n at gmail.com> wrote:
> I've been pondering for some time whether OSM should be hosting our own
> slippy maps or whether our role should be to *only* provide raw data.
> Leaving it to others to render and publish slippy maps etc.
We need a pretty front page map because that's a great show case for
OSM: we're making a map, and here it is!
What OSM also probably needs is some way to show case other OSM uses
from the front page. ie: if there are several other OSM layers in
existence that OSM doesn't need to host, but could easily direct users
The fun with having more layer types is that there are 100's of use
cases. Everyone will want a map to highlight something different --
that's the beauty of OSM, they can grab the data and do it -- and
show casing that is also a good thing to be doing.
The problem there is the resources needed to keep something like that
going are large. We currently have 1 server for a
render-on-the-fly-and-cache based layer which seems to be keeping it's
head mostly above water, and a server grinding away just keeping the
results of an entire planet of tiles for another 2(?) layers.
To expand this to 10 layers we need way more processing power -- and
it only gets worse with the number of users and layers expanding. If
OSM goes down this road then doing it for free will get tricky very
quickly. So I think OSM has 3 choices:
1) Provide a tool chain to make slippy maps easy to create - provide
a nice website taking users to all the wonderful layers people are
generating and doing the logistics for themselves
2) Have a server farm for generating tiles for every use case, and
charge for using it, or put ads on those maps it generates
3) Find a big pile of cash somewhere and do 2) for free
Personally I'd go with 1) and leave 2) as an interesting business
venture for someone :-)
More information about the talk