Classify devices functional role in the network. By evaluating the configured and operational states of devices, we can train models to categorize devices by role names that describe their function. This information, stored in the Role in Network API can be later used to help identify scope or priority of changes and recommendations.
How critical are the devices in a network? Does the function and configuration of device align with the perceived importance? The Place in Network Device Importance API creates a meaningful piece of per-device-metadata defining the criticality of a device based on its running features.
Starting with Network Profile, device configurations are processed through an extract engine to evaluate running configuration state. Feature extraction based on SME knowledge is used to create data points indicating design and intent of role for each device. For more information on the feature extraction and how these fit into the model, click here. An engineer creates classifiers that defines their network and training data for the model by selecting several devices that fit into each role. The Random Forest algorithm then builds out the model and predicts the role of devices based off of the training data. The result information is stored in the Place in Network API, where the role information can be queried via Network Profile deviceId or for all devices belonging to a Company Key.
How am I Spending My Time?
Configuration changes are associated with topics for aggregation. Configuration terms that are seen frequently together will gravitate towards the same topic. We can use this topic information to associate configuration snippets with other common configuration items and categorize the function of the bundle of configuration changes.
How am I Spending My Time? - Named Topics
Using the "Topic Mapper", we can associate keyword bundles discovered from the configuration topics and associate our own topic name. This is useful for trending configuration bundles that are frequently seen together, and categorizing the changes as a particular function or task.
What is a Topic? - Topic Keywords
Topics are derived from the frequency of terms showing up in a configuration snippet. The collection of these terms often associated with each other creates the topic. Some terms are more descriptive of the topic, and are assigned a higher "weight". This chart and table displays how particulary important a term was in defining the topic.
Recent Changes - Device Drilldown
These tables provide the last known change information detected in the environment, and devices that have changes over the selected time period. The graph link provides a relation overlay of the configurtation items that were removed (red) or added (green) and the parent objects to which the changes were seen. The DeviceId link will navigate to a deep dive for the device, where you can view the individual changes that were seen for the device and a frequency indicator of how rare the terms in the config snippets are when compared to the entire change database.