Links and Resources
1. Intro
For large data sets with many users simultaneously navigating across different locations, it would be impractical to always render the map on the server on demand, response times and server load would exceed all limitations.
The entire map area (like, sometimes, the world) is therefore split into square 'tiles' and pre-rendered for several fixed resolutions = scales. This implies the availability of fixed scales (typically around 19 from coarsest to finest) only, with the tiles held in caches in one or more places along the chain of delivery from server to client.
To appreciate the performance and to have a peek-under-the-hood of a WMTS, check out this example from OpenLayers!
A Web Map Tile Service (WMTS) is specified as a standard by the OGC and implemented in various architectures.
Another, more advanced development is the tiling, caching and delivery of vector tiles to the client. With this approach no pre-rendering is done on the server, but vector data are delivered (in square tiles, again organized into multiple levels - figure below) straight to the client where visualisation happens (see OSM and other examples below.
One concern by some actors is the potential for copying of original data by end users. If this needs to be avoided, web map tiles are the preferred approach.
