Google Maps Platform • Hacker News Board

Re: Google Maps Platform

#261

Post by apex18 » Thu, May 10, 2018, 7:24 PM UTC

jpdus wrote: post #24

> If you are currently using an API key, please read the email we sent you on May 1st, which provides a high-level estimate of whether your future bill will change. Did anyone actually get this mail? I am using a key and even got a receipt yesterday but did not get this mail about pricing changes (user from Germany). Edit: Wow. I am currently using Places Radarsearch for a side project which currently has a (generous…

If you were using the free tier of Places Nearby or Text Searches (aka radar) and staying just under the daily search limit, your previous yearly bill of $0 from Google will now be roughly:

1.7 million dollars

Re: Google Maps Platform

#262

Post by tj89 » Thu, May 10, 2018, 10:10 PM UTC

kirchhoff wrote: post #242

kinard wrote: post #12

WOW, my costs are going to go from about $320 per month to $3,750 per month. Time to find a different solution! GULP!

I'm putting together statements from those affected by the huge price increase to send to the Google Maps product team as a joint message. If you'd like to add yours, please email me at nick@instantstreetview.com

any luck on getting a response?

Re: Google Maps Platform

#263

Post by reustle » Wed, May 16, 2018, 1:51 PM UTC

jakecopp wrote: post #106

reustle wrote: post #33

Earlier quoted context omitted.

Agreed. I contribute to Google maps a lot and I've looked into contributing to open street maps but the options seemed pretty clunky. I recommend mapbox to all of my clients over Google maps.

Just click the edit button on https://www.openstreetmap.org/ , I find the interface quite nice. It's quite addictive! That being said, I don't know of a similarly easy to use editor on Android for simple things like adding opening hours or addresses.

Thanks, yeah I should have clarified. I would like to easily edit OSM on android :)

Re: Google Maps Platform

#264

Post by Axtrix » Thu, May 17, 2018, 12:53 PM UTC

zylepe wrote: post #253

Doctor_Fegg wrote: post #172

There's (justifiably) a lot of love for Mapbox in this thread as an alternative, but as remarked, their pricing isn't particularly cheap (why should it be, they sell on the quality of the product). So the other option is to self-host tools using OpenStreetMap data. Some pointers: * Raster tiled maps: the canonical instructions are at switch2osm.org; put it on commodity hardware such as Hetzner/OVH. * Vector maps: a l…

I recently did an evaluation of Valhalla, ORSM, and Graphhopper for routing based on openstreetmap data. Valhalla was nice in that it could preprocess and serve results with only a few GB of RAM, but it could only handle OSRM was super fast, especially for very short routes (>1k rps on a single core) but it requires 300+GB of RAM to preprocess the world and 64GB of RAM to serve direction requests which gets a bit too…

I'm going to do a similar test to evaluate some routing alternatives. How did you perform your tests?

Re: Google Maps Platform

#265

Post by zylepe » Sat, May 19, 2018, 7:51 PM UTC

Axtrix wrote: post #264

zylepe wrote: post #253

Earlier quoted context omitted.

I recently did an evaluation of Valhalla, ORSM, and Graphhopper for routing based on openstreetmap data. Valhalla was nice in that it could preprocess and serve results with only a few GB of RAM, but it could only handle OSRM was super fast, especially for very short routes (>1k rps on a single core) but it requires 300+GB of RAM to preprocess the world and 64GB of RAM to serve direction requests which gets a bit too…

I'm going to do a similar test to evaluate some routing alternatives. How did you perform your tests?

For routing performance I ran them each locally with instructions from their github repos and the OSM extracts for a single state or city from https://download.geofabrik.de then used https://github.com/codesenberg/bombardier to test latency/throughput with sample requests.

For resources to preprocess the world I ran Valhalla on my laptop and it preprocessed the world overnight. For graphhopper I kept running it on bigger and bigger digitalocean droplets until it just barely finished on their $640/month 64gb cpu-optimized droplet in about 12h. I didn’t even try osrm after reading that it needed about 300gb of ram.

Let me know if you have any better findings!