how to reduce Geoserver WMTS rendering time?

Question

I am publishing 100 Gb of aerial image til on GeoServer using pyramid technique. I have made GeoServer in a production environment according to GeoServer manual "Running in a production environment". I have prepared data using gdal_retile:

gdal_retile.py -s_srs EPSG:27700 -v -r bilinear -levels 4 -ps 500 500 -co 'TILED=YES' -co 'COMPRESS=JPEG' -targetDir OUTPUT *.jpg

Input jpg image details:

Driver: JPEG/JPEG JFIF Files: SS0096.jpg SS0096.jgw Size is 4000, 4000 Coordinate System is `' Origin = (200000.000000000000000,197000.000000000000000) Pixel Size = (0.250000000000000,-0.250000000000000) Metadata:
EXIF_BitsPerSample=8 8 8 EXIF_ColorSpace=65535 EXIF_Compression=1 EXIF_DateTime=2016:09:08 07:20:28 EXIF_Orientation=1
EXIF_PhotometricInterpretation=2 EXIF_PixelXDimension=4000
EXIF_PixelYDimension=4000 EXIF_PlanarConfiguration=1
EXIF_ResolutionUnit=2 EXIF_SamplesPerPixel=3 EXIF_Software=Adobe Photoshop CS2 Windows EXIF_XResolution=(72) EXIF_YResolution=(72) Image Structure Metadata: COMPRESSION=JPEG INTERLEAVE=PIXEL
SOURCE_COLOR_SPACE=YCbCr Corner Coordinates: Upper Left ( 200000.000, 197000.000) Lower Left ( 200000.000, 196000.000) Upper Right ( 201000.000, 197000.000) Lower Right ( 201000.000, 196000.000) Center ( 200500.000, 196500.000) Band 1 Block=4000x1 Type=Byte, ColorInterp=Red Overviews: 2000x2000, 1000x1000, 500x500, 160x160 Image Structure Metadata: COMPRESSION=JPEG Band 2 Block=4000x1 Type=Byte, ColorInterp=Green Overviews: 2000x2000, 1000x1000, 500x500, 160x160 Image Structure Metadata: COMPRESSION=JPEG Band 3 Block=4000x1 Type=Byte, ColorInterp=Blue Overviews: 2000x2000, 1000x1000, 500x500, 160x160 Image Structure Metadata: COMPRESSION=JPEG

Windows Server configuration is Intel Xeon E5, 2.20GHz (2 processor) and 32gb RAM. But when I am accessing this layer using WMTS service, it is taking a lot of time (around 20-30 minutes) to load completely.

Can anyone suggest me how to reduce rendering time?

Answer 1

The best advice is to follow Paul Ramsey's advice:

So, to sum up, your best format for image serving is:

• GeoTiff, so you can avoid proprietary image formats and nonsense, with
• JPEG compression, for visually fine results with much space savings, and YCBCR color, for even smaller size, and
• internal tiling, for fast access of random squares of data, and
• overviews, for fast access of zoomed out views of the data.

For gdal_translate he recommends the following options:

gdal_translate \
  -co COMPRESS=JPEG \
  -co PHOTOMETRIC=YCBCR \
  -co TILED=YES \
  5255C.tif 5255C_JPEG_YCBCR.tif

and for overviews:

gdaladdo \
  --config COMPRESS_OVERVIEW JPEG \
  --config PHOTOMETRIC_OVERVIEW YCBCR \
  --config INTERLEAVE_OVERVIEW PIXEL \
  -r average \
  5255C_JPEG_YCBCR.tif \
  2 4 8 16