CPU Activity: System Performance Metrics from a Multi-User Sun Workstation

Two collection epochs of system activity data gathered every 5 seconds from a Sun Sparcstation 20/712 in a multi-user university department. The dataset contains 22 attributes measuring memory, process, and system call activity, with the goal of predicting the portion of time CPUs run in user mode. The final dataset includes an equal number of observations from each collection period.

Use Cases

• Predict CPU user-mode time percentage based on memory transfer and system call rates.
• Model relationships between process run queue size and available memory or swap space.
• Analyze correlations between page fault types and system read/write activity.
• Develop benchmarks for system performance under varied user task loads.

Strengths

• Data was collected continuously with observations taken every 5 seconds.
• Includes 22 detailed performance metrics covering memory, process, and system call activity.
• Final dataset is balanced with equal numbers of observations from two separate collection occasions.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.
• Data originates from a specific hardware and environment, which may limit generalizability.

Provenance

Source

OpenML platform, originally collected from a Sun Sparcstation 20/712 in a university department.

Collection Method

System activity was gathered continuously every 5 seconds during two separate occasions.

Time Range

Specific dates are not provided.

Freshness

Last updated date is unknown.

Geography

Location is not specified, but the environment is a multi-user university department.