Understanding Metadata

Metadata Overview

Metadata refers to arbitrary information that can be associated with other data (monitored objects) to enhance their description. While some systems implement labels, Cisco Provider Connectivity Assurance (formerly Skylight performance analytics) utilizes a key–value pair approach.

The following diagram illustrates this difference:

Metadata enables efficient navigation, filtering, segmentation, and correlation across large volumes of items—such as tens or hundreds of thousands of monitored objects. It serves as a critical mechanism for isolating and analyzing information relevant to a specific context.

Examples of Metadata Categories

• Source / start / origin / probe
• Source type (e.g., VCX, nano)
• Destination / end / reflector
• Endpoint type (e.g., router, eNB)
• Geographic coordinates (latitude and longitude) of the object or endpoints — considered special metadata
• Region or network segment
• Customer
• Topology or path information (e.g., core and end router names) — considered special metadata
• Link type (fiber, microwave, etc.)
• Measurement type (telemetry, TWAMP, etc.)
• Class of Service (CoS) — Data, Voice, etc.
• Equipment vendor
• Configuration parameters
• Responsible department for object performance

Applications of Metadata

1. Filtering

Wherever the filter icon is available, metadata values can be used to refine the result set.

Key Filtering Notes:

• Filters are cumulative and use AND conditions—adding filters typically narrows the result set.
• Filters may be applied by a combination of multiple key–value pairs.
• Both inclusive filters (e.g., region = EAST + WEST) and exclusive filters (e.g., all regions except NORTH) are supported.

Filter Bar

The filter bar is available across Provider Connectivity Assurance, including dashboards, analysis, inventory, and reports. Users can type or scroll to select desired metadata categories.

Features:

• Saved Filters: Bookmark commonly used filter combinations for quick reuse.
  

• Starred Metadata: Embed selected metadata categories directly into dashboards for rapid filtering.
  

Filtering from Tables and Legends

You can select metadata from tables and line chart legends, then add them as filters.

2. Grouping

Some dashboard widgets provide grouping capabilities by metadata category.
Grouping consolidates performance metrics of all sessions sharing the same key–value pair, producing aggregated timeseries (charts) or singular values (tables or histograms). This approach consolidates large volumes of data into digestible insights via a dashboard, making it easier to identify potential issues.

3. Correlation

The Analysis application leverages metadata extensively to assist in determining potential root causes of incidents.

For more details, refer to the Analysis Distribution Widget.

4. Permission-Based Access Control (PBAC)

Metadata can be used to restrict user access to specific subsets of monitored objects by applying permissions at the metadata category level. Permissions are assigned to user groups.

Example:

• User Paul is limited to monitored objects where customer=MC.
• User Frank can access monitored objects where customer=BK.

Planning Considerations

Effective PBAC implementation requires upfront planning to partition monitored objects according to your operational needs.

Recommendation: Use fewer, larger segments over numerous small segments to maintain both performance and manageability.

Creating Metadata Permissions

Once metadata categories and object partitions are established:

1. Create a user group.

2. Assign one or more metadata permissions to that group.
   

3. Assign the group to one or more users.
   A user inherits all permissions from their assigned groups. For simplicity and maintainability, keep both the number of groups and the number of permissions per group minimal.

Note: Filters are impacted by PBAC settings—a user can only filter categories and values they are permitted to access.

Example:

• The customer category has values MC, BK, and 5G.
• User Frank has access to MC and BK and can create additional filters for either value.
• User Paul has access only to 5G and therefore cannot filter further within that category.

Restrictions and Best Practices

• Maintain an optimal metadata set—limit the number of categories to 50 for ease of management.
• Where possible, automate metadata updates using available APIs and system integrations.
• Metadata applied post-creation does not retroactively affect historical data. If you plan to use metadata permissions, ensure to partition your monitored objects in advance.
• Keep the number of permissions and user groups low for better performance and easier administration.
• A single user must not exceed 20 metadata permissions in total (across all assigned user groups).

Limitations

• Maximum field length: 255 characters per metadata value.
• Keys may only include alphanumeric characters (0-9, a-z) and underscores (_); no spaces permitted.
• Values use UTF-8 variable-length character encoding.
• All keys and values are stored in lowercase (e.g., Meta_Data is stored as meta_data).