Recommended configurations for optimal performance
To achieve optimal performance for publishing and navigating the user interface, the administrators need to understand the volume of data, hardware resources and user load. Doing so helps schedule publishing, exporting, and uploading of the data in an ideal way.
- Example harware setup: Windows Server 2019 Standard 64-bit; Intel Xeon CPU E5-2640 v4 @ 2.4GHz (2 CPU); 32 GB RAM; 2 SSD HDs: 100 GB and 300 GB.
- Java memory configuration: Use default settings described in Configuring JVM memory settings in LiveContent Central.
- binary-path-rollover: 2000
- Deployer worker queue: In deployer-conf.xml, uncomment
workers propertyand modifyqueueWorkerCountto -2 . - Database (PostgreSQL) settings:
- statement_timeout = 3700000 (default 0)
- max_connections= 1000 (default 100)
- shared_buffers= 128MB (default)
- Database connection pool settings: Use default settings described in Configuring the database connection pool.
Using the configuration described above, RWS has successfully published and navigated through various data sets. A simple dataset of 1 PMC, 2000 DMCs, 700 graphics (2GB uncompressed size) took 20 minutes to publish. A complex dataset of 17500 PMCs, 87500 DMCs, 700 graphics (2.5GB uncompressed size) took around an hour.
Please note that there is a link between the number of packages generated by Publish Service and Deployer's configuration for the workers property. If your data modules and images are only KBs in size and you publish many of them, keeping the default or increasing the settings of the binary-path-rollover and Deployer queueWorkerCount is recommended. However, if your data modules and images tend to be MBs in size and you publish many of them, the value of the binary-path-rollover and Deployer queueWorkerCount can be decreased.
The security software like virus scanning (e.g.,Windows Defender) has a direct impact on the speed of file operations. RWS recommends turning off Windows Defender Antivirus for optimum performance.