Heuristic Experiment in Dynamic Vector Tile Generation for Web Maps

_{For the keen eyes, I'm using the new Vivaldi web browser}

This experimental project is a visualizer of my Vector Tile Encoder as it dynamically generates vector tiles for web applications based on the Javascript API for ArcGIS. Specifically, this visualizer mimics the response of a VectorTileServer REST request emitted by a VectorTileLayer to dynamically search a data source and transform the result set into vector tile features.

The server application is based on Spring Boot and I highly recommend the Spring Initializer for a quickstart.

Fundamentally, the application consists of responding to 3 mapping requests:

• /arcgis/rest/services/{datasource}/VectorTileServer - This returns the vector tile metadata in JSON for the specified {datasource} such as the full extent of the datasource, min/max scales for visibility and the resolution and scale at different level of details.
• /arcgis/rest/services/{datasource}/VectorTileServer/styles - This returns the Mapbox GL style in JSON that defines what data to draw, the order to draw it in, and how to style the data when drawing it. Note - Not all the styles are implemented in this project.
• /arcgis/rest/services/{datasource}/VectorTileServer/tile/{z}/{x}/{y}.pbf - This returns an array of bytes consisting of the Protocol Buffer encoding of a tile based on the Vector Tile Specification for a specified datasource at a specifed level z and specified horizonal x and vertical y placements.

In Action

The easiest and most reliable way to reduplicate the visualization process in action is to use Docker.

• Start a machine named dev using the VirtualBox driver

docker-machine create --driver virtualbox --virtualbox-memory "1024" --virtualbox-cpu-count "1" dev

• Get the machine's IP

docker-machine ip dev

• Setup the docker environment variables

eval $(docker-machine env dev)

• Run the Vector Tile Server

docker run -it --rm -p 8080:8080 mraad/vector-tiles-boot

• Open a browser and navigate to http://IP-FROM-ABOVE:8080
• Shutdown the server by navigating to http://IP-FROM-ABOVE:8080/shutdown.html

The Data Source

As of Apr 8, this implementation relies on a static data source. Later project updates will have a dynamic data source. The static data source is a GZip text file containing the world administrative polygons in geographical coordinates. Each line in the text file is a polygon feature with a set of tab separated attributes and the geometry is in WKT format.

Building the Project

mvn clean package

Building The Docker Image

cd docker
cp ../target/vector-tiles-boot-0.0.2-SNAPSHOT.jar .
docker build\
 --force-rm=true\
 --ulimit nofile=65535:65535\
 --ulimit nproc=25059:25059\
 -t mraad/vector-tiles-boot .

Using Maven

Found this Maven plugin that allows me to control builds and deployments:

mvn package docker:build

Starting and stopping the container:

mvn docker:start

mvn docker:stop

See Also

• https://blogs.esri.com/esri/arcgis/2015/07/20/vector-tiles-preview/
• https://github.com/mapbox/tilejson-spec
• http://wiki.openstreetmap.org/wiki/Slippy_map_tilenames#Scala
• https://github.com/mapbox/awesome-vector-tiles
• https://www.mapbox.com/mapbox-gl-style-spec/
• http://www.diva-portal.org/smash/get/diva2:851452/FULLTEXT02
• https://github.com/mapbox/mercantile
• https://github.com/opensciencemap/vtm
• https://docs.spring.io/spring-boot/docs/current/reference/html/deployment-install.html
• https://hub.docker.com/r/gliderlabs/alpine/
• https://developer.atlassian.com/blog/2015/08/minimal-java-docker-containers/
• https://github.com/anapsix/docker-alpine-java

License

Apache License, version 2.0