Storage performance, particularly IOPS and latency, is critical for the stability of the UIM database and message routing. While specific IOPS numbers depend on your environment's scale (number of devices and monitors), we recommend the following best practices for sizing:
- Storage Type: We strongly recommend using high-performance SSD storage for the Primary Hub and the UIM Database server. For Medium to Large deployments, 300GB+ of SSD storage is recommended to handle queue backups during maintenance or outages.
- Database Performance: The database is the most IOPS-intensive component. We recommend a minimum of 64GB of dedicated RAM for the database server and low-latency storage to ensure rapid QOS data insertion and alarm processing.
- Latency: To maintain optimal performance, storage latency should be minimized. High disk latency can cause the nisqueue.db to grow and result in delayed alarms in the Operator Console.
- Environment Variation: Production environments should strictly follow the Large-Scale or Medium-Scale reference architectures. UAT and Development can often use smaller footprints but should maintain the same architectural separation (separate servers for DB, Hub, and OC) to accurately reflect production behavior.