Antonio,
Apologies for the slow response and thank you for the encouraging words.
Our
gitter, as you've seen, is a much more active venue for GeoTrellis discussion.
It looks like you've had your questions answered there, but for posterity, I'll link to the responses:
1.
>>>
TileLayerMetadata
Talking more generic, probably it’s smth you want to know as well; we store metadata in a single json file in a set of json files.
The entire metadata stored in attribute store consists of Header (contains information about where and what type of layer is stored), Metadata (tile layer metadata), KeyIndex (KeyIndex that is used to index layer), Schema (Avro schema used to prevent avro tiles corruption)
2
>>>
hey @antonio-rodriges we’re using TIFF to read data
GeoTiff allows to read images smartly and to load the consistently on a Spark cluster
internaly we operate with Arrays of data (:
the output format is Avro (in case of GeoTrellis avro layers) or geotiffs (COGS, some custom exports / etc)
3.
>>>
it’s not possible to perform operations on a compressed data in RDDs; otherwise for each operation you’ll have to uncompress chunks to apply function to compress back; which is very expensive.
on the avro side it uses avro deflate compression
if you want to persist everything as COGs / TIFFs you can choose a deflate compression level using java deflate compressor
4.
>>>
Antonio Rodriges @antonio-rodriges Feb 18 08:10
Hi, when GeoTrellis reads a layer from HDFS, does it read MapFile by MapFile sequenstially or GeoTrellis reads all MapFiles in parallel?
Grigory @pomadchin Feb 18 08:31
@antonio-rodriges it uses SequenceFileInputFormat or FilterMapFileInputFormat depending on the request
Were you able to get all of your questions answered? Don't hesitate to follow up if you have additional questions. Although, As you've seen, you're likely to get a faster response on Gitter.
Best,
Ross
