ArcGIS Server Performance Recommendations: https://enterprise.arcgis.com/en/server/latest/administer/linux/tuning-your-arcgis-server-site.htm
Modifications Made:
We use arcgis-server-8.gisdemo.net server. The command is time wget HTML query URL. The parameter Max Mosaic Image Count controls the number of values returned. In CRF, it is set to null. In MRF, it is set to 50. In CoG/netCDF-4 it is set to 20.
Product | Format | Storage | Time (s) | Note |
---|---|---|---|---|
CERES | CRF | S3 | 4.2 | |
CERES | CRF | EBS (=Local) | 3.8 | |
CERES | CRF | S3 | 5.0 | returns 8,760 points even if input parameter is 20. |
CERES | CRF | Local | 5.1 | returns 8,760 points even if input parameter is 20. |
CERES | MRF | S3 | 2.4 | Output has much less points (=50) than CRF (=8,760). |
CERES | CoG | S3 | 3.6 | Output has 20 points. 8,760 CoG files. |
CERES | netCDF-4 | S3 | 1.24 | Output has 20 points. 8,760 netCDF-4 files. |
Please adjust lambda timeout parameter in serverless.yml for MDCS script or data conversion. Timeout can generate incomplete Mosaic Dataset (MD) or MRF file.
Product | From | To | Time | Size | Note |
---|---|---|---|---|---|
CERES | Mosaic Dataset | CRF | 2h 45m | 14GB | |
CERES | netCDF-4 on S3 | MD on local DB | 6h | 200 rows | ArcGIS Pro 2.6 / 8,760 = (24 * 365) rows. |
CERES | netCDF-4 local | MD on local DB | 1h 4m 50s | 2.58GB | ArcGIS Pro 2.6 / 8,760 = (24 * 365) rows. |
CERES | AWS RDS Table | MD on local DB | 1h 42m 20s | ArcGIS Pro 2.6 Input table points to 8,760 CoG on S3 Parallel processing parameter = 4 | |
CERES | AWS RDS Table | MD on local DB | 46m 46s | ArcGIS Pro 2.6 Input table points to 8,760 netCDF-4 on S3 Parallel processing parameter = 4 | |
MERRA2 | 365 file list text file | Table | 38s | 24 * 365 rows | Miniconda 3.7 python and ls2rds.py script (sqlalchemy) |
MERRA2 | 1 file list text file | Table | 0.35s | 24 rows | Miniconda 3.7 python and ls2rds.py script (sqlalchemy) |
MERRA2 | AWS RDS Table | MD on RDS | 3m 14s | 2 rows (Cache size: 230K/file) | 2 NetCDF subset on S3 MDCS script on ArcGIS Server Linux 10.7.1 |
MERRA2 | AWS RDS Table | MD on RDS | 1m 40s ~ 2m 30s | 24 rows | 24 NetCDF subset on S3 MDCS script on ArcGIS Server Linux 10.7.1 |
MERRA2 | AWS RDS Table | MD on RDS | 3m 54s | 24 * 4 rows | 4 NetCDF files on S3 MDCS Script on ArcGIS Server Linux 10.7.1 |
MERRA2 | AWS RDS Table | MD on RDS | 3m 6s | 24 * 365 rows | 365 NetCDF files on S3 ArcGIS Pro on Windows Server (Adding Rasters only; no pixel cache; no multidim geoprocessing) |
MERRA2 | MD on RDS | MD on RDS | 1hr | 24 * 365 rows | 365 NetCDF files on S3 MDCS Calculate Statistics script on ArcGIS Server Linux 10.7.1 BMI is very quick. |
MERRA2 | MD on RDS | CRF on local disk | 25m 48s | 19.4GB | Using ArcGIS Pro 2.5.1 with the following options
|
MISR | netCDF-4 | netCDF-4 | 7m 30s | 454M | Subsetting 365 netCDF-4 files and merging with ncks from the downloaded files onArcGIS Server Linux 10.8. |
MISR | netCDF-4 | Mosaic on local disk | 19m 18s |
| |
MISR | Mosaic on local disk | CRF on local disk | 10hr 10m 53s | 2.4GB | Using ArcGIS Pro 2.6 with the following options
|
MOP03J | HDF-EOS5 | MRF | 0.065 | 24M | C++ (gdal_translate) / 1 file |
MOP03J | HDF-EOS5 | MRF | 0.365 | 24M | Python (mrf.py) / 1 file |
MOP03J | AWS RDS Table | MD on RDS | 2m 30s | 3 rows | 3 MRF files on S3 |
Sentinel-2 | AWS RDS Table | MD on RDS | 17m 50s | 789 rows |
|
TerraFusion | netCDF-4/HDF5 | OPeNDAP HTML form | 4m | Source file: 32.8G | AWS Open Data S3 on us-west-2 |
TerraFusion | netCDF-4/HDF5 S3 | OPeNDAP DDS only | 2m17s | Source file: 32.8G DDS: 199.78k | AWS Open Data S3 on us-west-2 wget |
TerraFusion | netCDF-4/HDF5 S3 | GeoTIFF on local disk | 109s | 25G | EC2 miniconda 3.7 gdal + pyresample & S3 on us-east-1 MODIS 2030x1354 subset |
TerraFusion | netCDF-4/HDF5 S3 | GeoTIFF on local disk | 157s | 32.8G | EC2 miniconda 3.8 gdal 3.0.2 + pyresample & S3 on us-west-2 MODIS 2030x1354 subset |
TerraFusion | netCDF-4/HDF5 local | GeoTIFF on local disk | 0.9s | 32.8G | EC2 miniconda 3.8 gdal 3.0.2 + pyresample MODIS 2030x1354 subset |
TerraFusion | netCDF-4/HDF5 S3 | GeoTIFF on local disk | 14.9s | 32.8G | EC2 miniconda 3.8 gdal 3.1.2 + pyresample & S3 on us-west-2 MOPITT 436x29 subset |
TerraFusion | THREDDS S3 | GeoTIFF on local disk | 604s | 32.8G | EC2 miniconda 3.8 pydap + pyresample & S3 on us-west-2 |
TerraFusion | THEDDS Local | GeoTIFF on local disk | 1s | 32.8G | TerraFusion file was copied into local drive. |
TerraFusion | THEDDS Local | netCDF4 on local disk | 30.7s | 32.8G | TerraFusion file was copied into local drive. |
TerraFusion | THEDDS Local | netCDF4 on local disk | 10s | 32.8G | TerraFusion file was copied into local drive. |
Product | From | To | Time | Size | Note |
---|---|---|---|---|---|
MERRA2 | ArcGIS Server | AWS RDS | 1.6s | Empty input table | Create MD input table using psql command line tool. |
MERRA2 | ArcGIS Server | ArcGIS Portal | 31.5s | 2 Rasters in MD | Upload service using upload.py arcpy script. |
MERRA2 | ArcGIS Server | ArcGIS Portal | 15.8s | 2 Rasters in MD | Delete service using portal.py arcpy script. |
MERRA2 | ArcGIS Server | AWS RDS | 23.5s | 2 Rasters in MD | Delete mosaic dataset using delete.py arcpy script. |
MERRA2 | ArcGIS Server | AWS RDS | 1.6s | 2 rows in input table | Delete input table using psql command line tool. |
Product | From | To | Time | Size | Note |
---|---|---|---|---|---|
TerraFusion | S3 | ArcGIS Server | 189Mbs/sec | 32G |