The scatterplot

The Scatterplot plots live data for each connected N|Solid process on two axes, using any two N|Solid metrics. Use it to see clusters of similar processes and to spot anomalous processes and outliers across a wide range of behaviors.

Open the Scatterplot from the navigation menu at the top left.

Changing axes

By default, the Scatterplot shows Heap Used versus CPU % for all processes. Click either axis to change its metric to any of the metrics collected by the N|Solid Runtime.

You can also change the scale of each axis from Linear (the default) to Logarithmic or Exponential to visually separate clusters at the low or high end of the data range.

Axis scale examples

Linear

A linear x-axis spreads the processes evenly over the data range.

Logarithmic

A logarithmic x-axis increases the resolution of lower values at the cost of resolution in higher values.

Exponential

An exponential x-axis increases the resolution of higher values at the cost of resolution in lower values.

Sorting and selecting processes

Use the SORT menu above the process list to group and sort processes by App Name, Dead Processes, CPU, Hostname, Memory, Tags, Uptime, Vulnerabilities, or Worker Threads.

Here the list is sorted by Memory (Heap Used):

Click a group header in the list to select those processes. Here the application "worker" is selected. Click Apply as Filter to move the selection into the global Filter Bar, so the filter stays in place as you move to other sections of the Console. Click Clear to clear the selection.

Visual identification of outliers

The Scatterplot is designed to make outliers visible across a wide range of metrics. On their own, metrics say little beyond "using 100% CPU or memory is bad." The Scatterplot's real-time view gives metrics context.

An outlier is a member of a set that isn't behaving like the rest. Not every outlier is a problem, but understanding outliers helps you understand how your application operates and catch issues early. An outlier can come from expected variability in your data, or it can indicate a rare bug or a memory leak. Here are a few examples of ways to slice your data to find potential issues.

Outlier examples

RSS Outlier

This comparison of Heap Used vs. RSS separates the purple dot on the far right, whose RSS use is anomalous compared to all the others and worth investigating.

GC Outlier

This comparison of Heap Used vs. the number of garbage collections identifies two processes whose GC behavior differs from the rest.

HTTP 1

This Scatterplot shows Heap Used versus the number of HTTP(S) requests each server has handled since startup. The dots on the far left aren't servers, so they have served 0 requests. The middle shows the heap memory used by the various server processes.

HTTP 2

This Scatterplot shows RSS versus the number of HTTP(S) requests each server has handled since startup. One of the purple "worker" processes stands out as an RSS outlier: it uses significantly more RSS to handle a similar number of requests.

Log Scale

This example uses different axis scales to push an outlier to a visual extreme.

Working with outliers

The Scatterplot has tools for watching or highlighting processes of interest. Hover over a process in the list to show a crosshair icon, then click it to track that process's dot with a tail. The tail shows whether the process stays an outlier or periodically moves back into the group.

To filter to a subset of processes, drag a box around them with your mouse. The Scatterplot process list is filtered to that set. You can then apply the selection to the global Filter Bar, for example to look at the historical metrics of these processes on the Dashboard.

After you click Apply as Filter, the Scatterplot selection becomes a global filter:

Open the Dashboard from the navigation menu. The filter still applies, so you can continue diagnosing the selected processes: