You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, each GOL contains a snapshot of OSM data, imported from a specific .osm.pbf file.
In order to get an updated version, users need to download a new .osm.pbf file (or apply .osc files to an existing one), and build a new GOL from that file. Fortunately, build is relatively fast, but it still less than ideal if frequent updates are required. An ideal scenario would be the ability to update the GOL with just the OSM data that changed.
Here is the process we have in mind:
GOL files are already designed to be patchable. In theory, we could accommodate minutely updates, but our initial design is for a process that can run on a daily, or possibly hourly basis (The update-processing software would require greater complexity to do its job in 60 seconds).
Instead of directly applying .osc change files to the GOL, a separate program first pre-processes each .osc file into a set of patch files (one for each tile in which features have changed) and a version manifest (a small file that indicates which tiles have changed).
Ideally, this would run on a central server, so downstream users can skip the build import (which is comparatively resource-intensive) and just download tiles and patches.
For data consumers, the workflow would then look like this:
to download the current version of the planet (or only the regions required). The compressed tiles are similar in size to data in osm-pbf format. The tiles are simply unzipped and written into the GOL, so this step happens essentially at download speed.
Using
update planet <tileset-url>
you can then periodically update your GOL to the most current state of the data. This should only take seconds if the GOL is already reasonably current (again, basically whatever time is required to download the patches).
Updating happens as a two-step process: First, the client-side engine downloads the version manifest (which indicates the tiles that changed) and marks those tiles in the user's GOL file as stale. It then downloads patch files that contain the changes and applies them to the tiles to bring them to the current version.
Updating can be eager (all stale tiles are immediately brought current by the update command) or lazy (patches are downloaded and applied to stale tiles only once they are actually required by a query).
The text was updated successfully, but these errors were encountered:
The update command will support incremental updates directly from .osc files. The process will likely be sufficiently fast to process minute-by-minute updates, even on low-end systems. GOLs that are updatable in this manner will need to be built with additional indexes, increasing required storage by about 40%.
This form of updating will be able to generate patch files, which can be downloaded by data consumers that use "regular" GOLs (no additional indexes needed; updates in seconds).
Currently, each GOL contains a snapshot of OSM data, imported from a specific
.osm.pbf
file.In order to get an updated version, users need to download a new
.osm.pbf
file (or apply.osc
files to an existing one), andbuild
a new GOL from that file. Fortunately,build
is relatively fast, but it still less than ideal if frequent updates are required. An ideal scenario would be the ability to update the GOL with just the OSM data that changed.Here is the process we have in mind:
GOL files are already designed to be patchable. In theory, we could accommodate minutely updates, but our initial design is for a process that can run on a daily, or possibly hourly basis (The update-processing software would require greater complexity to do its job in 60 seconds).
Instead of directly applying
.osc
change files to the GOL, a separate program first pre-processes each.osc
file into a set of patch files (one for each tile in which features have changed) and a version manifest (a small file that indicates which tiles have changed).Ideally, this would run on a central server, so downstream users can skip the
build
import (which is comparatively resource-intensive) and just download tiles and patches.For data consumers, the workflow would then look like this:
Instead of
you would use
to download the current version of the planet (or only the regions required). The compressed tiles are similar in size to data in osm-pbf format. The tiles are simply unzipped and written into the GOL, so this step happens essentially at download speed.
Using
you can then periodically update your GOL to the most current state of the data. This should only take seconds if the GOL is already reasonably current (again, basically whatever time is required to download the patches).
Updating happens as a two-step process: First, the client-side engine downloads the version manifest (which indicates the tiles that changed) and marks those tiles in the user's GOL file as stale. It then downloads patch files that contain the changes and applies them to the tiles to bring them to the current version.
Updating can be eager (all stale tiles are immediately brought current by the
update
command) or lazy (patches are downloaded and applied to stale tiles only once they are actually required by a query).The text was updated successfully, but these errors were encountered: