Introduction to Skylar AI

Skylar AI is a software services suite powered by artificial intelligence (AI) that is designed to automatically manage and anticipate IT incidents. Skylar AI reasons over telemetry and the stored knowledge of an organization to deliver accurate insights, recommendations, and predictions.

This chapter provides an overview of Skylar AI and the various components that use Skylar AI.

What is Skylar AI?

Autonomic IT leverages artificial intelligence (AI), automation, and data to intelligently self-manage an entire IT stack. Autonomic IT drives autonomous businesses with rapid decision-making, cost-optimized scalability, and innovative experiences that empower organizations to focus on core innovation. The Skylar AI platform, which includes Skylar Analytics and Skylar Advisor (coming soon), helps customers with their journey towards Autonomic IT.

Skylar One collects data and leverages Skylar AI to learn the patterns for a particular device metric over a period of time. Skylar uses the resulting data to build a device metric-specific model that is used to define a scope of expected behavior as well as anomalous data points.

Features of Skylar AI

Skylar AI is the engine that powers several different software components. The components in the Skylar family of services share the following characteristics:

Reactive. When something fails, Skylar AI tells you in plain language what happened and how to fix it with relevant context.
Predictive. Skylar AI alerts you in advance to an expected out-of-capacity condition.
Proactive. Skylar AI accurately answers any question asked of it with context drawn from company knowledge sources, such as bugs, support tickets, Knowledge Base articles, and Product Documentation, and recommends next steps.

Skylar AI integrates seamlessly with the Skylar One platform and other IT management tools. You can interact with Skylar AI through these familiar environments, where it enhances existing workflows with AI-driven insights and automation capabilities. Skylar AI can send you alerts and notifications, which can be customized to suit individual preferences or organizational needs. These alerts help you stay informed about potential issues, ongoing incidents, or opportunities for optimization.

Components of Skylar AI

The Skylar AI family of services includes the following components:

Skylar Analytics, an advanced reporting and custom analytics service that combines AI-powered analytics with deep data exploration and visualization.
Skylar Advisor, a proactive IT problem-solving advisory service powered by human-centered AI.

This manual covers Skylar AI. At the Product Documentation site, you can find documentation for Skylar Analytics and Skylar Advisor.

Data Analyzed by Skylar AI

The following image shows the flow of data into and out of Skylar One (formerly SL1) and the Skylar AI Engine:

The following list contains some of the types of data that Skylar One can send to the Skylar AI engine, where the data is analyzed and used by Skylar Analytics and Skylar Advisor:

Alert and event logs
Asset data
Availability data collected by Skylar One
Business Service health, availability, and risk metrics from Skylar One
Class-Based Quality-of-Service (CBQoS) metadata and CBQoS time series data
Device category and device group data
DCM(+R) relationships
Dynamic Application mapping and performance data, and journals
Email round trip data
Event data, including event policy name
Interface billing metrics and metadata
Internal Collection processes, collection services, and CBQoS
Journal Dynamic Application data
Metadata for web content, SOAP/XML transaction, and domain name monitors
Process and service data
Skylar One Agent data, including Gen 1 (Skylar One Distributed Environment) and Gen 3 (Skylar One Extended Architecture) agents
Thresholds
Topology data for L2, L3, CDP, LLDP, and ad-hoc relationships between devices

Configuring Skylar AI

These instructions are only for on-premises configurations of Skylar AI. The ScienceLogic SRE team performs these steps for SaaS configurations of Skylar AI.

Before you can start using Skylar AI components, you will need to perform the following configurations in Skylar One to enable the export of data from Skylar One to Skylar:

Create a Service Connection
Enable Skylar Analytics for one or more organizations

After you perform these configurations, you can access Skylar Analytics, Skylar Advisor, and other key Skylar AI components from the Skylar AI page () in Skylar One.

For information about setting up users, user groups, and user roles, see Configuring Skylar AI System Settings.

ScienceLogic strongly recommends that you always use the most recent Skylar One and AP2 releases in conjunction with the most recent Skylar AI release. Using the most recent releases will ensure that your Skylar AI system has access to the latest datasets and features. For more information, see the Skylar One Platform and AP2 Release Notes.

Creating a Service Connection

If you are using AP2 Mochi or later with your Skylar One system, you can create a service connection for the Skylar AI engine on the Service Connections page (Manage > Service Connections) in Skylar One. ScienceLogic strongly recommends that you upgrade to Mochi or later. For more information, see the AP2 Mochi release notes.

The service connection enables communication between your Skylar One system and Skylar AI. This process replaces the Running the Skylar Management Tool process in previous releases of Skylar Analytics and Skylar One.

To create a Skylar AI Engine service connection:

In Skylar One, go to the Service Connections page (Manage > Service Connections).
Click Add Service Connection and select Skylar AI Engine. The Create Skylar AI Engine Credential window appears.
Complete the following fields:

Name. Type a name for the new service connection.
API Key. Add the access token for Skylar AI, which you can generate on the Access Tokens page in Skylar Settings (Instances > Access Tokens). For more information, see Creating Access Tokens for Users.
Skylar AI Engine URL. Add the URL for your Skylar AI system.

Click Save. The service connection is added to the Service Connections page, and a modal displays a link to the Organizations page, where you can enable Skylar Analytics for one or more organizations. See the following procedure for more information.
Refresh or reload the browser to add all updates to Skylar One.

Newer releases of Skylar One include a Status and Status Updated column, along with a Service Check column that displays a Run Test button for "Skylar AI Engine" service connection types. Click Run Test to run a script to check the status of the Skylar AI connection and display the results in a modal.

Enabling Skylar AI for One or More Organizations

You will need to select one or more organizations in Skylar One that will share data with Skylar AI. This data will come from all of the devices in a selected organization. By default, the Skylar AI features are disabled.

You can see which organizations are currently sending data to Skylar AI by going to the Organizations page (Registry > Accounts > Organizations) and looking at the Skylar AI Status column for the organizations.

To enable Skylar AI with Skylar One organizations:

In Skylar One, go to the Organizations page (Registry > Accounts > Organizations) and click the check box for one or more organizations.
In the Select Action drop-down, select Send Data from Selected Orgs to Skylar AI and click Go to start sending data about the selected organizations to Skylar AI. The Skylar AI Status column for the selected organizations changes to Enabled.
If you want to override the hard-coded default options for exporting metadata to Skylar AI, you can add the JSON values in the Skylar Options (JSON) text field on the Behavior Settings page (System > Settings > Behavior). the values entered must be in the same JSON structure as those that appear in the config.py file:
```
DEFAULT_OPTIONS = { "metadata": { "intervals": {"snapshot": 60,
"cleared_events": 5}, "snapshot": { "batch_sizes": { "asset_basic":
500, "asset_ip_config": 10000, "perf_index_label": 5000, "device_
config": 20000, }, }, "cleared_events": { "query_range": 60, }, } }
```

You can use the "Skylar: Failed" event policy in the "Skylar One Default Internal Events" PowerPack to raise an event in Skylar One if Skylar fails. If you installed Skylar One version 12.5.1 from an ISO file, this PowerPack will be included with the installation. If you are upgrading to Skylar One version 12.5.1, you will need to download the most recent "Skylar One Default Internal Events" PowerPack from the PowerPacks page on the ScienceLogic Support Center (Skylar One > PowerPacks) and then install the PowerPack.

A generic, internal alert definition is generated in Skylar One to indicate whenever a Skylar AI connection is successful or encounters a problem.

For Older Versions: Running the Skylar Management Tool

If you are using a version of AP2 before Mochi, you will need to set up Skylar AI with the steps below for the Skylar SL1 Management Tool instead of the Service Connections page in Skylar One. ScienceLogic strongly recommends that you upgrade to Mochi. For more information, see the AP2 Mochi release notes.

The Management Tool configures Skylar One data and Skylar One processes, and it starts monitoring the Skylar connection and configuration. The script is named sl-otelcol-mgmt.py, and it is included in the sl-otelcol RPM package.

To run the Skylar SL1 Management Tool:

Use the following command to run the Management script on the Database Server (a Skylar One Central Database or a Skylar One Data Engine):

sudo sl-otelcol-mgmt.py -vv skylar --skylar-all --skylar-endpoint "<URL_for_skylar_system>" --skylar-api-key "<skylar-access-token>" --ap2-feature-flags

where:
- <URL_for_skylar_system> is the URL for your Skylar AI system
- <skylar-access-token> is the access token for Skylar AI, which you can generate on the Access Tokens tab of the Skylar Settings page. For more information, see Creating Access Tokens for Users.

This command configures the OpenTelemetry Collector, restarts services that export data, and checks that connectivity to the supplied endpoints is healthy.

You can also use the following configuration options if needed:

--verify-cert false. Allows users in on-premises environments to connect to Skylar AI using self-signed certificates.
--ca-bundle /<path>/bundle.pem. Allows users to specify a path to a .pem file and assign it to the REQUESTS_CA_BUNDLE environment variable.
--skylar-disable. Stops all Skylar AI exports and services. This flag performs the same operations as the pause command (see step 3, below) and also removes any Skylar AI pages from the Skylar One user interface.

If you have already run setup before and are not changing the connection details, you do not need to include --skylar-endpoint "<URL_for_skylar_system>" --skylar-api-key "<skylar-access-token>".

In addition, --ap2-feature-flags is only needed the first time you install Skylar AI.

After successfully running the script, on the System Logs page (System > Monitor > System Logs), you will see "Info" messages for each configuration change (filter on sl-otelcol-mgmt). You will also see "Major" system log messages whenever connectivity fails for the Skylar endpoint or the OpenTelemetry Collector.

After data streams into the Data Visualization dashboards, and other Skylar AI components, they will populate with data. Please note that this process might take several minutes.

If you have run the setup script before, run the following command to enable Skylar AI and make sure that everything is working as expected:

sudo sl-otelcol-mgmt.py -vv skylar --skylar-all
If you need to pause Skylar AI, run the following command:

sudo sl-otelcol-mgmt.py -vv skylar

Pausing sets all Skylar AI toggle fields to disabled; restarts the event engine and data pull services to reflect the changed configuration; stops Skylar One managed services such as the Metadata Exporter, Alerts Poller, and sl-otel-mgmt.timer; and stops and disables the sl-otelcol systemd service.
To check the status of the installation, run the following command:

sudo sl-otelcol-mgmt.py -vv status

You should look for the following messages in the output:

---------- checking feature toggles

SL_EXPORT_EVENTS = False

SL_EXPORT_METRICS = True

SL_EXPORT_CONFIG = True

---------- checking services

sl-otelcol is enabled and running

---------- checking connectivity

checking: Skylar endpoint is healthy

checking: local OTELCOL endpoint is healthy

If you need to turn off the Skylar AI connection, run the following command:

sudo sl-otelcol-mgmt.py -vv skylar --skip-status-service
Go to the previous procedure to specify the organizations you want to use for exporting data to Skylar.

Enabling Skylar AI Event Policies

In addition, the Predictive Alerting and Anomaly Detection components of Skylar Analytics require the "Skylar Analytics Event Policies" PowerPack. This PowerPack collects the Skylar One event policies from the "Skylar - Predictive events" and "Skylar One: Skylar Anomaly Score Event Monitoring" PowerPacks.

Older versions of this PowerPack were named "Skylar Predictive Analysis".

To install the "Skylar Analytics Event Policies" PowerPack:

Search for and download the PowerPack from the PowerPacks page at the ScienceLogic Support Center (Skylar One > PowerPacks, login required). Alternatively, you can use the link provided by ScienceLogic, if applicable.
In Skylar One, go to the PowerPacks page (System > Manage > PowerPacks), click Actions, and then click Import PowerPack.
Browse and select the downloaded PowerPack and click Import.
On the next screen, click Install and, when prompted for confirmation, click OK.
To confirm that the PowerPack was installed properly by go to the Event Policies page (Events > Event Policies) and type the word "predictive" into the Name search field. You should see a number of "Predictive Alerting" event policies.

For information about how to use these components, see the following chapters:

Logging In to the Skylar AI User Interface

You can access Skylar AI components from a link in Skylar One, or if you know the URL of your Skylar AI system, you can go directly to that location instead of using Skylar One.

From Skylar One, go to the Skylar AI page () and click the Visit button for the Skylar AI component you want to use, such as Skylar Analytics or Skylar Advisor. If you are not currently logged in to Skylar AI, the Skylar AI sign-in page appears.

Clicking the Visit button for Skylar Predictive Alerting and Skylar Anomaly Detection opens new pages for those components in Skylar One.
If you need to log in to Skylar AI, type your email address and password and click Continue. The Skylar AI landing page appears.
Click the name of the Skylar AI component you want to use, such as Analytics or Skylar Settings .

For more information about the Skylar Settings page, see Configuring Skylar AI System Settings.