Using the default config from https://raw.githubusercontent.com/graphhopper/graphhopper/master/config-example.yml and texas-latest.osm.pbf, I am finding it takes a long time to calculate 90 minute isochrones (approximately 90 seconds). As I have a substantial number of isochrones to crunch through, I am curious if there are any additional optimizations.
I think I am using CH, as per the graph-cache/properties file
90 minutes / 75 miles are my max use case. This distance is unsupported in the vast majority (or all?) of available 3rd party isochrone creators.
Server setup: 16GB VM with 4 vcores on a recent Dell XPS15 w/ 64GB RAM (Fedora33 host, Ubuntu 18.04 guest)
java -Xmx8192M -Xms4096M -Ddw.graphhopper.datareader.file=texas-latest.osm.pbf -jar *.jar server config-example.yml
What can I do reduce my run time from ~90 seconds per API call?