RNFdev Web Space and band Width
Dan Putler
putler at sauder.ubc.ca
Fri Sep 22 12:50:19 EDT 2006
Hi Dave,
> 1. I'll chat with Frank about the next step.
Good, but I think we need some data before this becomes important. It
gets at questions of what is the right release strategy. Should we
put Ottawa up ASAP, or should we wait until we have an entire
province (or the entire country) before releasing any data?
> 2. I will also ask Scott and Frank about mirrior issues and how we can
> spread the love.
This sounds right. My guess is that this will be a "peak load"
problem. The first few months there will be lots of hits, and then
things will really slow down.
> 3. If we are having 300 downloads a month then perhaps there will be
> more support out there to mirror and such. At least it is a start. If
> the need is bigger then maybe the Telescience approach would work.
My demand estimates are a wild guess, so this is hard to say.
> 4. I'm glad to see thoughts about WMS and WFS. Makes sense realy. This
> is something to ask the mapserver folks as I'm sure there is a recipe.
It seems like this is the way of the future, so it should always be
up on the list. I also think PostGIS is the way of the future of open
source GIS, but I'm not sure what the implications of this are since
there are multiple ways of sucking up shapefiles in PostGIS.
> 5. About the size of the file. Maybe we can cut that down a bit. What
> if we ditch the geography of it? How much space do we save?... I
> know it
> sounds weird. What I'm thinking is that we create a DBF table that can
> be easily linked (one to one) with the RNF file. That way it is a two
> step process. First download the RNF. Then download our RNF upgrade
> and
> provide a means to append. Going this route would also benefit in that
> if the RNF file changes we can find a way that we can always link our
> DBF to it, instead of us needing to maintain the geography. I'm am
> thinking of linking to polyID or something like that. And we can start
> with a standard test to see if the poly ID's and their roadnames have
> changed. Just some version control thoughts. How does that sound in
> principle? This might be similar to building a schema to link two DBF
> files? Then we house the schema (or what ever I should be calling it)
> and not the raw geography.
I'm with you on this, but there are problems. If the NGD_IDs were
unique, we would be in great shape since a merge on a single key
could be made. However, as I indicate in my polygon creation recipe,
the NGD_IDs aren't unique. It may be possible to create a unique key
from the existing fields in the RNF, but I'm not sure.
> I'm researching how to extract data from PDF a little more. I will
> maybe
> try Toronto GTA area and some other southern ontario spots. I figure
> this area is good for testing and building because:
> A. Lots of potential users
> B. It's probably going to be the biggest sluggish portion.
Yeah, the whole damn country revolves around Toronto, so it makes
sense as a plan.
Dan
More information about the Can_rnf
mailing list