[OSM-dev] Mass imports (TIGER and AND)

[Dave Hansen]

On Tue, 2007-08-28 at 20:53 +0200, Martijn van Oosterhout wrote:
> On 8/28/07, Ray Booysen <raybooysen at rjb.za.net> wrote:
> > Will not moving some of the work out to seperate clients not solve this
> > problem?  With testing you're doing requests from one machine whereas
> > the server should be able to handle multiple requests from multiple
> > clients better.
> 
> The biggest problem with parallelisation is the feedback: when you
> create a node you need to store the node number because later a
> segment will use it. And the id of that segment will later be used to
> create a way. So you need to think of a method to feed this data back
> to a central store.
> 
> That's not to say it can't be done, but it's a lot more work than just
> spawning a few extra processes and giving each a part of the file.

With the TIGER data that isn't a problem.  We can simply upload multiple
counties in parallel.  They don't share ids, and no ways or segments
span their boundaries.  

-- Dave