Laurent Grégoire

Do we have an estimate of the amount of data to store? Why not using a standard database such as PostGreSQL? Data could be archived and stripped after several months...

Just for the record, here are some statistics we gathered on some [OTP optimization experiment](https://github.com/opentripplanner/OpenTripPlanner/wiki/Experiment:-Memory-usage-statistics), based on OSM data: ``` Netherlands Paris (IDF) NY state ----------------------------------------------------------------------- Number of street edges...

On many systems an opened file take a non negligeable amount of memory ([around 1K for linux](https://github.com/torvalds/linux/blob/master/fs/file_table.c#L319)), and usually non swappable. So leaving around many opened files can be an...

In term of export format, AFAIU OpenLR could encode this kind of data, since the format allow multiple shortest paths sharing common edges. So you could encode a shortest path...

There exists [regional extracts](http://zverik.osm.rambler.ru/gps/files/extracts/index.html) that can simplify things where only a small region is needed (GPX files in each extracts are stored by lat/lon bins). Also lots of those files...

I think we are leaving a bit the original scope of this issue, which was about distributed servers (pool storage / traffic engine). Maybe we should create another issue to...

Maybe we could merge the concept of "distributed servers" (at least part of it, about geographical balancing) with the concept of "tiling". Both do have more or less the same...

To help anonymization, there is probably no need to store the contextual "path" information for each segment, ie the preceding / following speed profile. This may not be enough, but...

I think the solution depends on the context of use: internal ID vs an exchange format with the external world. In the following I'm only talking about an internal ID...

Topological information can be useful when resolving a GPS trace to the proper segments: you can use the fact that the vehicle at the source of the trace has to...