Dynatrace provider

The Dynatrace provider collects metrics from Dynatrace and makes them available to Akamas.

This provider includes support for several technologies. In any case, custom queries can be defined to gather the desired metrics.

Supported versions

Dynatrace SaaS/Managed version 1.187 or later

Supported component types:

• Kubernetes and Docker

• Web Application

• Ubuntu-16.04, Rhel-7.6

• java-openjdk-8, java-openjdk-11, java-openjdk-17

• java-ibm-j9vm-6, java-ibm-j9vm-8, java-eclipse-openj9-11

Refer to Dynatrace provider metrics mapping to see how component-types metrics are extracted by this provider.

Prerequisites

This section provides the minimum requirements that you should match before using the Prometheus provider.

• Dynatrace SaaS/Managed version 1.187 or later

• A valid Dynatrace license

• Dynatrace OneAgent installed on the servers where the Dynatrace entities to be monitored are running

• Connectivity between Akamas and the Dynatrace server on port 443

• A Dynatrace API token with the privileges described here.

Dynatrace Token

The Dynatrace provider needs a Dynatrace API token with the following privileges:

• metrics.read (Read metrics)

• entities.read (Read entities and tags)

• DataExport (Access problem and event feed, metrics, and topology)

• ReadSyntheticData (Read synthetic monitors, locations, and nodes)

• DataImport (Data ingest, e.g.: metrics and events). This permission is used to inform Dynatrace about configuration changes.

To generate an API Token for your Dynatrace installation you can follow these steps.

Component configuration

To instruct Akamas from which Dynatrace entities (e.g. Workloads, Services, Process Groups) metrics should be collected you can some specific properties on components.

Different strategies can be used to map Dynatrace entities to Akamas components:

• By id

• By name

• By tags

• By Kubernetes properties

By id

You can map a component to a Dynatrace entity by leveraging the unique id of the entity, which you should put under the id property in the component. This strategy is best used for long-lived instances whose ID does not change during the optimization such as Hosts, Process Groups, or Services.

Here is an example of how to setup host monitoring via id:

name: My Host
properties:
 dynatrace:
  id: HOST-12345YUAB1

You can find the id of a Dynatrace entity by looking at the URL of a Dynatrace dashboard relative to the entity. Watch out that the "host" key is valid only for Linux components, other components (e.g. the JVM) must drill down into the host entities to get the PROCESS_GROUP_INSTANCE or PROCESS_GROUP id.

By name

You can map a component to a Dynatrace entity by leveraging the entity’s display name. This strategy is similar to the map by id but provides a more friendly way to identify the mapped entity. Beware that id multiple entities in your Dynatrace installation share the same name they will all be mapped to the same component. The Dynatrace display name should be put under the name property in the component definition:

name: MyComponent
properties:
 dynatrace:
  name: host-1

By tags

You can map a component to a Dynatrace entity by leveraging Dynatrace tags that match the entity, tags which you should put under the tags property in the component definition.

If multiple tags are specified, instances matching any of the specified tags will be selected.

This sample configuration maps to the component all Dynatrace entities with tag environment: test or [AWS]dynatrace-monitored: true

name: MyComponent
properties:
 dynatrace:
  tags:
     environment: test
     [AWS]dynatrace-monitored: true

Dynatrace supports both key-value and key-only tags. Key-only tags can be specified as Key-value tags with an empty value as in the following example

name: MyComponent
properties:
 dynatrace:
  tags:
     myKeyOnlyTag: ""

By Kubernetes properties

By leveraging dedicated properties, you can map a component to a Dynatrace entity referring to a Kubernetes cluster (e.g., a Pod or a Container).

Container

To properly identify the set of containers to be mapped, you can specify the following properties. Any container matching all the properties will be mapped to the component.

Akamas property	Dynatrace property	Location
namespace	Kubernetes namespace	Container dashboard
containerName	Kubernetes container name	Container dashboard
basePodName	Kubernetes base pod name	Container dashboard

You can retrieve all the information to setup the properties on the top of the Dynatrace container dashboard.

The following example shows how to map a component to a container running in Kubernetes:

dynatrace:
  type: CONTAINER_GROUP_INSTANCE
  kubernetes:
    namespace: boutique
    containerName: server
    basePodName: ak-frontend-*

Pod

To properly identify the set of pods to be mapped, you can specify the following properties. Any pod matching all the properties will be mapped to the component.

Akamas property	Dynatrace property	Location
state	State	Pod dashboard
namespace	Namespace	Pod dashboard
workload	Workload	Pod dashboard

If you need to narrow your pod selection further you can also specify a set of tags as described in the by tags. Note that tags for Kubernetes resources are called Labels in the Dynatrace dashboard.

Labels are specified as key-value in the Akamas configuration. In Dynatrace’s dashboard key and value are separated by a column (:)

Example

The following example shows how to map a component to a pod running in Kubernetes:

dynatrace:
  type: CLOUD_APPLICATION_INSTANCE
  namePrefix: ak-frontend-
  kubernetes:
    labels:
      workload: ak-frontend
      product: hipstershop

Container, Pod, or Workload?

Please note, that when you are mapping components to Kubernetes entities the property type is required to instruct Akamas on which type of entity you want to map. Dynatrace maps Kubernetes entities to the following types:

Kubernetes type	Dynatrace type
Docker container	CONTAINER_GROUP_INSTANCE
Pod	CLOUD_APPLICATION_INSTANCE
Workload	CLOUD_APPLICATION
Namespace	CLOUD_APPLICATION_NAMESPACE
Cluster	KUBERNETES_CLUSTER

Improve component mapping with type

You can improve the matching of components with Dynatrace by adding a type property in the component definition, this property will help the provider match only those Dynatrace entities of the given type.

name: MyComponent
properties:
 dynatrace:
  type: SERVICE     # the type helps the mapping by tags by filtering down entities that are only services
  tags:
     environment: test
     "[AWS]dynatrace-monitored": true

The type of an entity can be retrieved from the URL of the entity’s dashboard

Available entity types can be retrieved, from your Dynatrace instance, with the following command:

curl 'https://<Your Dynatrace host>/api/v2/entityTypes/?pageSize=500' \
  --header 'Authorization: Api-Token <API-TOKEN>'

Mapping multiple entities in one component

In some circumstances, you might want to map multiple Dyantrace entities (e.g. a set of hosts) to the same Akamas component and import aggregated metrics.

This can be easily done by using tags. If Akamas detects that multiple entities have been mapped to the same component it will try to aggregate metrics; some metrics, however, can not be automatically aggregated.

To force aggregation on all available metrics you can add the mergeable: trueproperty to the component under the Dynatrace element.

name: MyComponent
properties:
 dynatrace:
  mergeable: true
  tags:
     environment: test
     [AWS]dynatrace-monitored: true