The Node Performance panel shows the performance metrics for all nodes in a managed server. You can view node performance data for the past year, month, week, last 24 hours, or a custom date range.
Use the following options to display the data in this panel:
- To show only a selected node or nodes, use the View By list. By default, data for all nodes is shown.
- To get the node performance forecast for the next 90 days, use the Show Forecast toggle.
- To add or remove metrics for a node from the graph, click the node label.
The panel includes the following information:
- Service Time Latency
- Service time latency in milliseconds. Service time latency measures how quickly the node processes and delivers information.
- OPS
- The server operations per second per node.
- FPGA Load
- The percentage of the maximum workload that the field programmable gate arrays (FPGAs) can support. When the FPGA load reaches 100%, the system is operating at full capacity and cannot process requests any faster.
Because the system contains multiple FPGAs, the reported FPGA load at any given time reflects the load on the busiest FPGA. The load is influenced by the time the FPGAs spend actively processing requests and the time spent waiting for required resources before processing can begin.
- CPU Load
- The percentage of the maximum workload the CPU can support at its current clock speed. To reduce power consumption, the server dynamically adjusts the CPU clock speed based on demand. As a result, during periods of low CPU utilization, the reported CPU load percentage might appear higher than it would if the CPU were running at full speed.
When CPU utilization is high (greater than 80%), the CPU load shows operation at the maximum clock speed. CPU load is influenced by multiple factors, including the volume and type of workloads applied to the system, such as NFS exports, SMB shares, NDMP operations, and file system checks.