Skip to main content
Migration Notice
We're migrating documentation from the old portal into this one. Some things may look a little different or out of place in the meantime — we know, and we're working to get it right. If something's unclear or doesn't look right, let us know.

Job Trends

The Job Trends page shows throughput and performance charts for a single copy job. To open it, click View Long-term Trends in the job's Job #N Details flyout.

Use it to understand how a job performed over its lifetime:

  • How fast objects were copied.
  • Whether threads were fully used.
  • Where latency spikes occurred.

Charts are available for both completed and in-progress jobs.

Overview​

The copy engine samples performance data every 10 seconds while the job runs. The charts grow in real time during a job and are preserved after the job completes.

The page refreshes automatically every 30 seconds. The line under the charts shows:

  • The number of samples.
  • The seconds until the next poll ("refresh in Ns").
  • The query time in milliseconds.

The Refresh button at the top fetches new data at once.

Latest Stats Summary​

A row of summary cards above the charts shows the most recent sampled values:

  • Current objects per second.
  • Current MB/sec.
  • Current active thread count.
  • The latest average GET and PUT latency readings.

For completed jobs, these values reflect the final sample taken before the job ended.

Chart Panels​

Each chart panel is shown only when data is available for that metric.

Copy Rate and Throughput​

Copy Rate - Objects / Second and Throughput - MB / Second are two charts. They plot objects per second and megabytes per second over the job's run time. Use them to see the copy speed profile across the job lifetime.

  • A flat line at zero can mean the job paused or was throttled.
  • A steep ramp at the start is normal while threads warm up.

Cumulative Objects Copied and Active Threads​

This chart combines two lines:

  • A running total of objects successfully copied, plotted over time. Its slope shows the effective throughput.
  • The active thread count on a second axis. It shows how many copy threads were active at each sample point.

A section of the objects line that goes flat means no objects were being completed. Look for matching spikes in the latency chart to understand why.

The chart does not draw the thread maximum. Compare the active threads line with the thread setting the job ran with. If active threads stay well below it, the job may be I/O-bound, or the source or destination may be throttling requests.

GET and PUT Latency​

This chart plots the average and p95 latency over time, in milliseconds. GET is the read from the source. PUT is the write to the vault.

  • Spikes in GET latency indicate pressure or slowness on the source storage system.
  • Spikes in PUT latency indicate pressure on the vault destination or network congestion.

Workflows​

Identify periods of slow throughput in a completed job​

  1. Open the Throughput chart.
  2. Look for sections where objects per second dropped significantly below the peak rate.
  3. Compare those time periods against the Latency chart. High latency during the same window explains the throughput drop.
  4. Compare against Active Threads. If threads also dropped during that window, a bottleneck reduced parallelism.

Determine if the thread pool was fully utilized​

  1. Open the Cumulative Objects Copied & Active Threads chart.
  2. Check whether the Active Threads line reached the job's thread setting at any point. The maximum is not drawn on the chart.
  3. If active threads never reached the maximum, the job was likely I/O-bound rather than CPU-bound. The source or destination may have been throttling requests.

Monitor a running job in real time​

  1. Open the Job Trends page while the job is running.
  2. Watch the Throughput and Cumulative Objects charts extend as new samples arrive. The page refreshes every 30 seconds.
  3. If the Cumulative Objects line goes flat unexpectedly, check Copy Jobs - Live View & History to confirm that the job is still in running state.

Tips​

  • Charts are populated only while the job is or was running. If you open Job Trends for a job that was canceled immediately after starting, the charts may be empty or show only a single sample.
  • A gap in the Cumulative Objects chart does not always mean the job paused. It can indicate a batch of objects that took longer than one sample interval to complete, which is common for very large objects.
  • Use the p95 latency line rather than the average to detect tail latency problems. If p95 latency is high but the average is low, a small number of slow objects are distorting performance.
  • For very long-running jobs, the charts may compress horizontally to fit the full time range. Use the zoom controls, if available, to inspect a specific time window.