Understanding the magic of Metadata

Metadata is arbitrary information that users can add/tag against other data (or in our case monitored objects) to better describe it. Some systems use labels to achieve this (i.e. Skylight orchestrator), but Skylight analytics uses key/value pairs.

The following diagram is probably the best way to describe the difference:

Metadata allows for easier navigation, filtering, segmentation, and correlation of large volumes of items - like 10, 20, or 500 thousand monitored objects. It becomes an imperative tool for slicing and dicing your data to find what's relevant to your current context.

Example Metadata Categories

• Source / start / origin / probe
• Source type (i.e. VCX, nano)
• Destination / end / reflector
• Endpoint type (i.e. router, eNB)
• Geo coordinates (latitude and longitude) of your object or its origin and end points - considered "special" metadata
• Region or network segment
• Customer
• Topology or path information (core and end router names) - considered "special" metadata
• Link type (fibre, microwave, etc...)
• Measurement type (telemetry, TWAMP, etc..)
• Class of service (COS) - Data, Voice, etc...
• Equipment vendor
• Config parameters
• Department responsible for object performance

Uses for Metadata

Filtering

Wherever you see a filter icon, that's your chance to put all of your hard work of enriching your data to use!

A few notes about filters:

• These filters stack using AND conditions, meaning more filters and fewer objects.
• You can select to filter on the presence of the metadata key or a combination of selected metadata values.
• You can create inclusive (region EAST + WEST) or exclusive filters (all regions but NORTH)

Filter bar

The filter bar appears in many contexts (dashboards, analysis, inventory, reports).
Type or scroll to find the metadata category you're interested in using for a filter.

Leverage our saved filter feature to bookmark frequently used:

Also make us of our Starred metadata feature to author a category selection into a dashboard for quick access:

Filtering from tables and legends

Tables and line chart legends allow for multi-selecting metadata keys and adding them as filters:

Grouping

Many dashboard widgets will give users the option to group by a metadata category. In these cases, performance for all of the sessions with the same key value will be used together to produce an aggregated time series (chart) or singular value (tables/histograms). This is a great way to roll up large amounts of data into some digestible pieces from a dashboard to spot problems.

Correlating

The Analysis application uses metadata heavily to suggest potential root causes of incidents being investigated.

Learn more here - Analysis Distribution Widget

Permission Based Access Control (PBAC)

Metadata can also be used to restrict access to a subset of monitored objects by defining permissions on metadata categories. Metadata permissions are specified at the user group level.

Example:

User Paul may only see monitored objects of customer=MC, while user Frank has access to monitored objects of customer=BK.

If you consider using metadata permissions, you have to plan ahead!
To use PBAC, you need to partition or segment your monitored objects in advance. Think of metadata categories and values that are intuitive to your needs.

When you partition your monitored objects for PBAC, you want to favor fewer large segments over many small segments, both to maintain query performance and permission maintainability.
If you have many user groups with a dozen permissions each, you should re-think your partitioning strategy.

Create Metadata Permissions

Once you have provisioned the desired categories and partitioned your monitored objects, you can create a user group and assign one or more metadata permissions to it.

A user group can be assigned to any user of your tenant. A user inherits all metadata permissions of all assigned groups. While this provides great flexibility, it is recommended to keep the number of groups as well as the number of permissions per group small.

Example:
The metadata category customer has three potential values: MC, BK, and 5G.
User Frank has permission to access MC and BK. Frank could define a filter to further reduce the result to monitored objects of customer MC.

User Paul has only access to 5G. Paul cannot specify any filters for the customer category, since he is already restricted to a single value.

Restrictions / Words of Wisdom

• Don't muddy the water too much... having hundreds of different keys is not necessarily going to enrich your experience (and going to make keeping it all up-to-date a hassle). Analytics will limit you to 50 categories.
• Automate where you can - Skylight analytics has a great selection of APIs that can be used to programmatically keep metadata up to date through simple integrations to the respective IT systems.
• When adding new metadata categories, they won't apply to historical data. If you plan to use metadata permissions, make sure to partition your monitored objects in advance.
• Use as few metadata permissions and groups as possible, to maximize maintainability and query performance.
• A single user must not have more than a total of 20 metadata permissions across all his or her assigned user groups.

Limitations

Be aware of these limitations:

• Metadata keys can only use alphanumeric characters (0-9, a-z) and the underscore (_), with no spaces.
• Metadata values use UTF-8 variable-length character encoding.
• Keys and values are stored in lowercase. Any capitalization is automatically changed to lowercase. For example, Meta_Data is stored as meta_data.