pelias VS planetiler

https://pelias.io is another self hostable one. In my experience it performs better than nominatim

The stack I describe in the post is only for map tiles - Map tiles are a good fit for CDNs because the input space is small (just Z/X/Y coordinates on a square grid) and thus very cacheable.

Geocoding is a very different problem because the input space - human language - is much, much larger, and answering queries quickly to support features like autocomplete really requires a server with hot data in memory.

One of my favorite projects in this space is Pelias https://pelias.io which is an open source auto-completing geocoder based on OSM plus other open data. It's backed by a great team that also runs a business: Geocode Earth https://geocode.earth

Pelias is a self-hostable option.

If anyone comes across this looking for an alternative, we can help at Geocode Earth (https://geocode.earth).

We're a small independent company that has been working on geocoding since 2013, first as part of Mapzen(https://mapzen.com), and then with our own self-funded business after Mapzen shut down at the start of 2018.

Our core software, the Pelias Geocoder (https://pelias.io) is open source, and ironically we understand it to be the primary geocoder in use by Positionstack.

We're biased of course, but we think that as the primary authors of the open source project, and with a high-volume service that has had _zero_ major outagee, we're a great option for anyone who needs forward or reverse geocoding, addresss parsing, place or address autocomplete, etc.

Here is why you should contribute to OSM even though there are major players profiting from it:

OSM is big enough and good enough that all the tech giants (except Google) would do better to start with OSM and improve it to meet their needs than to start a new, completely proprietary map from scratch.

That means that we are in an amazing place where in addition to the substantial volunteer OSM community, there are contributions from Amazon, Apple, Facebook, and hundreds or thousands of smaller companies all coming back in to a single global map that everyone can use. It's very worth it to do work that strengthens OSM, as it increases the number of companies that will use it, and possibly contribute back, rather than doing work the world at large won't benefit from.

P.S. As a disclaimer, I am co-founder of Geocode Earth (https://geocode.earth) a small business that does indeed profit from OSM (and other open) data. We also contribute back both through OSM contributions and by releasing our core software as the Pelias geocoder (https://pelias.io)

Pelias uses ElasticSearch to merge these data sets. Yes, it is too much data to stuff into OSM, but basic address information wouldn't bloat OSM too much, imho.

Worth mentioning this project (https://github.com/onthegomap/planetiler) that lets you create osm mbtiles and pmtiles pretty easy!

For a self-hosted vector tile stack you can have a look into https://github.com/onthegomap/planetiler I found it very easy to get started and when you know the other stacks it is also very fast to create these vector tiles even for planet-scale.

(note, that I'm not affiliated with them, but they use some source code from us for the efficient import and also contributed to GraphHopper, but this did not influence my experience ;) )

> I wonder why so many seem to be moving away from raster tiles to vector data.

The flexibility of styling. And you can easily serve customers that need different default languages. This makes maps also more accessible for countries without Latin alphabet.

I used https://github.com/onthegomap/planetiler and https://download.geofabrik.de, maybe it helps.

I spent a while optimizing sqlite inserts for planetiler, this is what I came up with:

https://github.com/onthegomap/planetiler/blob/db0ab02263baaa...

It batches inserts into bulk statements and is able to do writes in the 500k+ per second range, and reads are 300-400k/s using those settings.

Checkout https://github.com/onthegomap/planetiler.

Super easy way to generate a MBTiles, which you can then serve directly, or further convert to PMTiles, which can be used to host vector tiles for client-side rendering using MapLibre (or other renderers).

Raster tiles are a lot harder because you have to generate them on the server, and that's a lot more resource intensive.

I built planetiler (https://github.com/onthegomap/planetiler) for this purpose. The output up to z14 is ~80gb and depending on how big of a machine you have it takes from 30 minutes up to a few hours - no DB required, just java or docker. If you are only going to z11-12, it should be quite a bit faster/smaller.

Brandon from Protomaps is also helping add pmtiles output natively to planetiler, so you won't need a conversion step afterwards!

It's not raster directly, but you could use planetiler ( https://github.com/onthegomap/planetiler ) to build a full planet vector map . Then you could use something like TileServer-GL to server the vector map with a style. TileServer-gl ( https://github.com/maptiler/tileserver-gl) would provide a raster source that displays in the style you set on your vector map.

I recently left Twitter after 9 years, most recently serving as tech lead for the knowledge graph group (was 45 people). I helped apply the KG to drive a large portion of Twitter’s revenue and new product launches. In my spare time I do data visualization and web mapping, most recently https://github.com/onthegomap/planetiler