1. Overview

Collections are predefined filters for segments of your environment. They’re centrally defined, and they’re used in rules and views across the product.

Collections are used to:

  • Scope rules to target specific resources in your environment. For example, you might create a vulnerability rule that applies to all container images in an app called sock-shop. The rule might reference collectionA, which specifies sock-shop** in the image resource filter.

  • Partition views. Collections provide a convenient way to browse data from related resources.

  • Enforce which views specific users and groups can see. Collections can control access to data on a need-to-know basis. These are known as assigned collections.

Collections are created with pattern matching expressions that are evaluated against attributes such as image name, container name, host name, labels, function name, namespace, and more.

2. Partitioning views

While a single Console manages data from Defenders spread across all hosts, collections let you segment that data into different views based on attributes.

Collections are useful when you have large container deployments with multiple teams working on multiple apps all in the same environment. For example, you might have a Kubernetes cluster that runs a shopping app, a travel app, and an expenses app. Different teams might be responsible for the development and operation of each app. An internal tools team might be responsible for the travel and expenses app, while a product team runs the shopping app.

Selecting a collection reduces the scope displayed in Console to just the relevant resources. For example, the developer for the travel app only cares about vulnerabilities in the images that make up the travel app. All other vulnerabilities are just noise. Collections help focus the data.

3. Scoping rules

The scope of a rule is defined by referencing the relevant collections. Collections offer a centralized way to create and manage scope settings across the product. Collections make it easy to consistently reuse scope settings across policies. Policy tables give you a clear picture of what resources are being targeted in your rules.

collections policy table scopes

When creating new rules, you can either select from a list of previously defined collections, or create a new one. By default, Prisma Cloud sets a rule’s scope to the All collection, which captures all resources in the environment.

collections rule scope

Collections cannot be deleted as long as they’re being used by a rule. This mechanism ensures that rules are never left unscoped. Click on a specific collection to see how it’s being used.

collections in use

4. Importing and exporting rules

Rules can be exported from one Console and imported into another Console. When importing rules, any associated collections are also imported and created.

  • If the imported rule uses a collection that doesn’t exist in Console, the collection is automatically created.

  • If the imported rule uses collection with a name that already exists, but with a different scope, the collection is created with the following name and description:

    • Name: <policyType> - <ruleName> <collectionName>

    • Description: Automatically generated collection for an imported rule/entity

  • If the imported rule uses a collection that already exists, and a matching scope, the existing collection is used as-is.

5. Creating collections

You can create as many collections as you like. Collections cannot be nested.

Prisma Cloud ships with a built-in set called All that is not editable. The All collection contains all objects in the system. It is effectively the same as creating a collection manually and setting a wildcard (*) for each resource type (e.g., containers, images, hosts, labels, etc).

Collections can be created in Manage > Collections and Tags > Collections. Alternatively, collections can be created directly from a new rule dialog when you’re setting the rule’s scope. When creating collections from a new rule dialog, Prisma Cloud automatically disables any irrelevant scope fields. When selecting previously defined collections in a rule’s scope field, any improperly scoped collections are hidden from display. For example, you can’t select a collection that specifies serverless functions in a container runtime rule.

By default, new collections set a wildcard for each resource, effectively capturing all resources in the system. Customize the relevant fields to capture some segment of the universe of resources.

The labels field supports Docker labels, Kubernetes pod template labels, Kubernetes namespace labels, Kubernetes deployment labels, AWS tags, osDistro:<name> (for hosts), and osVersion:<version> (also for hosts). To use Kubernetes namespace and deployment labels, enable the following setting when deploying Defenders: Manage > Defenders > Deploy > DaemonSet > Collect Deployment and Namespace labels.

You cannot have collections that specify both containers and images. You must leave a wildcard in one of the fields, or else the collection won’t be applied correctly. If you want to create collections that apply to both a container and an image, create two separate collections. The first collection should only include the container name, the second should only include the image name. Filtering on both collections at the same time will yield the desired result.
Filtering by cloud account ID for Azure Container Instances isn’t currently supported.

To create a new collection:

  1. Open Console.

  2. Go to Manage > Collections and Tags > Collections.

  3. Click Add collection.

  4. In the Create a new collection dialog, enter a name, description, and then specify a filter to target specific resources.

    For example, create a collection named Raspberry images that shows all raspberry images in the fruit namespace. Pick a color for easy visibility and differentiation.

    The following collection selects all images that start with the string raspberry. You can also create collections that exclude resources. For more information on syntax that can be used in the filter fields (e.g., containers, images, hosts, etc), see Rule ordering and pattern matching.

    collections specify filter
  5. Click Save.

6. Selecting a collection

Collections filter data in the Monitor section of Console.

When a collection (or multiple collections) are selected, only the objects that match the filter are shown in those views. When a collection is selected, it remains selected for all views until it is explicitly disabled.

To select a collection, go to any view under Monitor. In the Collections drop-down list in the top right of the view, select a collection. In the following screenshot, the view is filtered based on the collection named google images, which shows all images that contain the string google_containers.

collections 792004

When multiple collections are selected, the effective scope is the union of each individual query.

Individual filters on each collection aren’t applicable to all views. For example, a collection created with only functions won’t include any resources when viewing hosts results. Similarly, a collection created with hosts won’t filter images by hosts when viewing image results.