Analysis Distribution Widget

The Distribution widget is the highlight in Analysis. It takes your categorical data (metadata and Per Packet Intel), aggregates against the values found in the dataset under analysis to produce an interactive distribution showing:

• Relative prevalence - what value is more common - shown via the size of the circle
• Aggregate performance - what's the average performance of all the objects (session) or rows (capture) for each value - shown via the bar and number

This is incredible powerful for quickly identifying outliers that might be indicative of the root cause of the issue being investigated.

Metric Selection

The distribution is driven off of a single metric. Users can choose the metric from the list metrics included within the scope of the view. If you're missing a metric, add it from the top toolbar and update your dataset.

Maximized Mode

When a widget is on the left hand side of the screen, it is said to be maximized. It is in this mode that all of the widget's features are available. Widgets on the right hand side of the screen are said to be minimized, but still over special capabilies! More on that below.

More Metric Info

Hover over any value and get more information!

• Range of values seen (min - avg - max) - this helps you understand the variance within the dataset for that category value
• Percentage of dataset with that value

Filtering

Analysis offers the ability to quickly add inclusion or exclusion filters via the distribution widget.
It's simple to use:

1. Select the filter mode as either Include or Exclude
2. Click on one or more category values (values can be within the same or different categories)
   – other widgets may react to this selection (for example the table and topology widget) while others will be gated on the next step (trends)
   – Exclude mode will react immediately to the section of a value by removing it from the list, while Include mode simply highlights the selected values
3. Notice the contextual options that appear at the top of the widget (a filter, a line chart, and an X).
   – Clicking the filter will promote the selection to the analysis filter bar
   – Clicking the line chart will quick plot (more on that below)
   – Clicking on the X will dismiss your distributions selections

Quick Plot a Reference Line

Selecting one or more category values from the distribution widget bring up the contextual options described above. One such option appears as a line chart, and will add a reference line based on those selected categories to the Trend or timeseries widget.

• the reference line will query Cisco Provider Connectivity Assurance (formerly Skylight performance analytics) for the aggregated time series data - it will not be limited to the cached data set being used by the analysis.
• selecting more than one value will result in the reference line being a combo of both (OR is the same category, AND if different categories
• you can remove the reference line by hovering over the row in the Trend legend and hitting the X

Graph Correlation

Category values are listed in order ascending order according to the worst aggregated performance (to promote the worst offenders). The graph correlation feature helps to visualize the relationship between categories since they aren't necessarily aligned like the Topology widget.

Categories Used and Column Ordering

Users to make transient (unsaved) changes to the analysis View, including on a per widget basis. The Distribution widget allows for the hiding of certain categories, as well as re-arranging the column order.

Minimized Mode

While minimized, the distribution widget will showcase the relative prevalence as bars, with the option to hover over bar sections for information on the category value it represents and the aggregated performance that value experienced. The list of categories is ordered from least number of different values to most,