I am running a GeoServer instance in docker and using that to render raster tiles but am finding the rendering to be very slow. I have tried adding the rasters to the Data Store as a single GeoTIFF, as an Image Mosaic from a dir of multiple rasters, and an Image Pyramid made from a raster with
gdal_retile.py, but they all seem to be very slow. I have found that the size (GBs) of the file doesnt seem to affect the speed of rendering speed, but is more to do with the number of pixels in the raster, which makes sense. I am going to have multiple raster files which collectively will cover the whole of the UK and will total hundreds of GBs. My question is really suggestions on how to improve the rendering speed of rendering from Geoserver.
I am running on docker on an EC2 instance, so one solution is likely to be simply launch an EC2 instance with more resources. Is there anything else I can do? Perhaps a way to make Geoserver render with slightly less accuracy or preprocessing the rasters to make it easier for Geoserver. Or even using Postgres as a data source rather than the GeoTIFFs directly?