Performance Analytics enables you to track, aggregate, and visualize key performance indicators for any specific time-period. Performance Analytics in
DX Operational Intelligenceallows you to:
- Detect performance bottlenecks and anomalies.
- Compare multiple metrics for the same devices, interfaces, networks.
- Compare multiple metrics across multiple entities.
- Compare uni-variate metrics across single and multiple entities.
- Compare multivariate metrics across multiple entities.
Entitycan be one of the following types:
- Device - For the network device, the interfaces can be selected as a context using the device itself.
Performance Analytics provides a historical analysis of multiple metrics and its machine learning based anomaly detection helps identify issues proactively. You can access Performance Analytics in two ways:
Access Performance Analytics
Follow these steps:
- Log in toDX Operational Intelligence, and click on from the left-navigation pane.
- Click the button.TheSelect contextpage appears.
- To view the performance charts incontext of an Alarm, use the following steps:
- Click theAlarmsicon on the navigation panel.TheService alarmspage appears.
- Click on an alarm from the list of alarms.The row expands and displays theAlarmdetails,Affected metrics,Impacted services,TopologyandAnnotation.
- Navigate toAffected metrictab.The performance chart for the selected metrics appears.
- From theAffected metrictab, click onCompare Metricsbelow the chart.ThePerformance Analyticspage is launched in context of the selected entity and alarm affected metric.
The following illustration depicts how you can access Performance Analytics:
The metrics can be ingested into Performance Analytics from the source products listed in Compatibility Matrix. The metric store processes raw metrics along with perform aggregation and anomaly detection. For more information on metric management, see Metric Management, and for more information on anomaly detection, see Anomaly Detection.
Performance Analytics User Interface
When you launch Performance Analytics page from the navigation pane for the first time, the page would be empty and you would be required to select the context and metrics. If you already have a saved view with context and metrics, then the Performance Analytics page opens with the default saved view. When you navigate from the Alarm Analytics page in the context of an alarm, the Performance Analytics page opens with the corresponding charts of the alarm.
Step 1: Select a Context
Select a context, perform the following steps:
- From thePerformance Analyticspage, click the button.TheSelect contextpage appears.
- Set the entity context by selecting entities from the available list of entities. Use theFilterfield to search for the entities.FilterYou can use the GlobalFilterfield to perform a text search across all the columns, and you can perform a more context based search using the contextFilterfield.Search for Interfaces:
States are calculated when a tenant has at least one service created. If the tenant has no services created, then all the entities are displayed as Normal.
- In the column search Filter, search forType: Devicesto get the results for associated interfaces with the router or switch.
- In the column search Filter, search forEntity Namesto get the results for associated interfaces with the router or switch.
- If you know the Interface name already, then you can search the specific interface name using the global filter.
- (Optional) You can sort theEntity namescolumn in ascending and descending order:
- By default, the entities are sorted in ascending order .
- To sort a column in descending order, click the column name till the descending order icon appear.
- To view only the selected entities, enable theDisplay ONLY selectedoption.
- ClickDoneat the top-left of the page.ThePerformance Analyticspage is set in the context of the selected entities. ForEntities, the page title is the name of the first entity and the number in the brackets is the count of the remaining selected entities.If you do not select any entity, thePerformance Analyticspage shows a blank page. Use the
After you selected the required entity context, you can proceed with the following steps:
Step 2: Select the Metrics
- From the selected entity page, click onMetricsoption.The list of available Metrics and Configuration Items (CIs) for the selected entities appear. The hierarchy that builds up depends on the selected context. If the selected context entity has more CIs associated to it, then the hierarchy builds up in such a way that you view the list of CIs and the associated metrics of the CIs.
- Select anEntityand click the drop-down arrow to view the associatedConfiguration Items.The list of associated CIs appear.
- Click the drop-down arrow to view the associatedMetrics Family.
- Click the drop-down arrow to view the associatedMetricsof the CIs. You can also clickFilterat the parent level to view the associated child entities.
- The hierarchies may vary from one source product to another.
- In the context search for an alarm or interface, sometimes you might not see the interface in context on theAvailable metricseditor, so filtering on that interface shows that interface and the selected metrics which is being plotted.
- While working on the Application context of APM sources, the APM entities that you select might not display the associated hierarchies. You must select the corresponding APM agent, then you can see the exact hierarchies and metrics.
- (Optional) Enable theDisplay Selected Metricsoption to view only the metrics that you have selected.
- Click theAdd Metric (+)button to add performance chart.The performance charts for the selected metrics appear on the Performance Analytics page.
Step 3: Metric Charts Views
This section describes the following chart views that are displayed based on data available:
- Timeline Series Probability Bands
- Timeline Series (Line) Charts
Timeline Series Probability Bandsare shown when there is a single metric selected, and for that metric anomaly detection is also configured. The
Timeline Series (Line) Chartsare shown when there is more than one entity being plotted on the same chart or the selected metric does not have anomaly detection configured.
If you have selected more than one Configuration Item or aggregation type as an entity, a timeline series chart is plotted. Click the legend to view the respective series in the chart.
If only a single Configuration Item (CI) is selected from the metric type family, a chart with probability bands is plotted. Based on the duration and frequency of the data that is captured the chart displays a Solid Green Highway Band, a Hash-Patterned Green Highway Band, or an Intermittent Pattern. You interpret the prediction bands to understand the performance of the metric.
Probability Band Charts: Interpret the Green Highway Band
When data for the metric is too large, a time slider is available below the chart. Use the time slider to select a time range and view details for a selected period. Three zones display the different prediction intervals present in the historical data. Scan the chart to identify points in time where an alarm was triggered, and where analytic data falls outside of the established norm for a component on your system. For more information on anomaly detection, see Anomaly Detection.
- You might experience a data break while using the Green Highway Bands for CA UIM and DX PM source products.This will be handled in a future rollout.
- The following metrics does not generate Metric Charts:
- DX APM - current capacity (bytes), amount of space used (bytes)
- DX PM - availability
A legend displays the different elements of the chart: You can switch the display of the different zones (rare, probable, common, and center) using the checkbox in the chart legend. For example, you can hide the display of the common and probable zones so it is easier to view high severity (rare) activity.
The chart combines information that is designed to help you quickly review and interpret activity for a metric in your environment. The various components of the chart are as follows:
- Actual DataActual data is illustrated as a black solid line (no alarms) or a colored solid line (alarm state) on the chart. The line is calculated from actual data for the metric. By comparing actual data and the median value, you can quickly see variations for the metric in your environment. When an alert occurs, the black line changes to a colored solid line that matches the alert severity and displays icons with the matching severity.For example, red for critical anomaly alarm, orange for major anomaly alarm, and yellow for minor anomaly alarm
- Mean Value for the MetricThe mean value or average for your metric is illustrated as a gray line on the chart. The mean value is displayed when there is not enough historical data.
- Center Value for the MetricThe center line is the middle of the range or median for the data and is shown as a green dotted line. The zones closest to the center line are closest to the most common measurements for the data.
- Common DataCommon zone data tracks closest to the center line, or the norm for your metric, and displays as a dark green band. The analytics calculations place the common zone as one percentile above or below the norm.
- Probable DataProbable zone data is illustrated as a green band on the chart. The analytics calculations place the probable zone two percentile points above or below the norm.
- Rare DataRare zone data is illustrated as a light green band on the chart. The analytic place the rare zone three percentile points above or below the norm, and signal a metric behavior outside the normal range.
- You can also view the chart data for an instant when you move your mouse over the chart to display dynamic information in a popup. For example, when you pan over the zones, each data point in time is summarized in a popup:
Hash-Patterned Green Highway
- When limited data is available for a metric initially, the Green Highway chart displays a hash-patterned band and represents recently observed performance of the metric, for example, in cases where a metric is newly added.
- The alerts that are produced are based on comparison between the most recently captured data (last few hours) and the data from longer timescales (days). As DX Operational Intelligence captures more data for a metric, the Green Highway chart switches to displaying from hashed to a solid pattern. Timescales for such a switch vary based on how often a given metric is collected and forecasts are calculated. For metrics that are reported to DX Operational Intelligence once per hour or more often, solid Green Highways appear within a week.The following image is an example of a hash-patterned Green Highway:
- Intermittent Hash-PatternWhen the number of data points are enough for the Solid Green Highways, then there is a switch from the Hash-Pattern to Solid Green Highways that is when you can see the Intermittent Hash-Pattern. The following image is an example of an intermittent pattern that occurs when there is a mismatch in the time of the data collection and the occurrence:
Solid Green Highway
- Explore Green Highway charts, which are a summary of past metric behavior at specific stages of your business cycle. These charts provide not only a qualitative description of when your system is the most or least active, but a quantitative description of what values for each metric are typical, and a forecast of where new values may appear.
- Produce anomaly alerts that are derived from the Green Highways. Because the Green Highways describe the likelihood of metric values, they also describe what values would be anomalous. DX Operational Intelligence uses Green Highways to:
- Generate statistical anomaly alerts for both individual extreme values
- Generate a series of values which are individually not unusual, but anomalous when observed together
Performance Analytics Page
This section describes the Performance Analytics page that appears after you have selected devices:
- Context Selection– Using this button, you can add or remove entities for which available metric charts are to be generated. Use theFilterfield to search for the entities.
- Aggregation Interval –Use the aggregation interval drop-down to specify the interval at which data aggregation takes place. If you selectRawoption from this drop-down, line charts are plotted in the metric charts and you cannot synchronize the metric charts. By default,Rawoption is selected. If you select a time range within 3 days, then the charts are plotted on raw metrics. If you select a time range beyond 3 days, then the charts are plotted on 15 minute aggregates and daily aggregates. For more information, see Metric Management.
- Grid View– Use the buttons to specify the number of metric charts to be shown in each row.
- Save View- Use this button to save a view. Once you select some entities and metrics, you can save your selections along with the metrics units as a view. So, when you refer back to your Performance Analytics page, you can see the saved view with metrics and metric units. For more information on the list of actions that you can perform onSave View, see Manage Views.In the Saved Views, if you have charts built on top of third-party metrics, the deselection of the charts will not be available.
- Metrics– Use this button to view the list of available metrics for the selected entities. Metrics are grouped based on the metric type. Configuration Items that belong to the same metric type are nested into the respective entity host name.
- Time Slider –Use this ellipse button to synchronize metric charts and view the synchronized timeline for a selected metric chart or all charts.
- Chart View– Selecting the metric creates a chart in the Charts panel. If only a single CI is selected from the metric type, a chart with probability bands is plotted. If you select more than one CI or multiple aggregation types are set to the entity, a line series chart is plotted.
You can synchronize metric charts and view the synchronized timeline for a selected metric chart or all charts.
Synchronize Metric Charts
Follow these steps:
- Select the required entities in thePerformance Analyticspage.
- Select the metrics from theAvailable Metricspalette.The metric charts are displayed.
- Select anAggregation Intervalfrom the drop-down.
- To synchronize a chart, switch on the synchronize toggle from the ellipsis button of a chart.The metric chart is synchronized according to the specified aggregation level. Use the Re-sync all time sliders option to synchronize all metric charts.
- Synchronization is possible only on time-based interval data, and not on raw data. By default, charts are synced and the default time interval is 10 minutes.
- If there are pre-existing entities with metric units, then these metrics would not be available in your current view after the current SaaS rollout. The metric units are available only for the newly created views. For more information, see Manage Views.
Click the date/time picker icon to select the duration for which you want to view the performance analytics for the selected entities. By default, the duration is one week. When you are accessing the Performance Analytics page from the navigation panel, then the default duration is one week.
Filter by Time Range
When you access the performance analytics page in the context of an affected alarm metric, then the default duration is calculated by (-8 hours + 1 hour) from the alarm creation time. the calendar is updated to match the last updated time of the affected alarm. This range is 8 hours before the last alarm update to one hour after the last alarm update.
You can use the date/time picker filter to set the duration for your search, that were created for a specific time period such as 1 hour, 24 hours, 1 week and so on. Performance Analytics also provides you with a
Customoption, which enables you to pick a particular start date/time and a particular end date/time to help you narrow down the metrics that you are looking for.
If you select a time range within 3 days, then the charts are plotted on raw metrics. If you select a time range beyond 3 days, then the charts are plotted on 15 minute aggregates and daily aggregates. For more information, see Metric Management.