CloudKitty Concepts ====================== This page provides the definitions for the concepts used in CloudKitty. It is recommended that you get familiar with them. Rating ------ It is the process of assigning a `value` to the consumption of computing resources. CloudKitty uses the concepts of services, which are rated. Therefore, one can configure services to be collected/monitored, and then through its processes, we can assign monetary values to the service consumption. The `value` assigned can be used to represent a monetary value in any currency. However, CloudKitty has no native module to execute conversions and apply any currency rate. The process to map/link a `value` to a real monetary charge is up to operators when configuring CloudKitty. Modules ------- Modules define the rating processes that are enabled. To get to know more about the rating modules, one should check `rating modules`_ . .. _rating modules: ../user/rating/index.html Services -------- Services define the metrics that are collected in a storage backend, and that are then rated by CloudKitty. Services need to be defined via API to be processed later by the rating modules, and configured in the collectors to be captured. Services are configured to be collected in the ``metrics.yml`` file. More information about service creation can be found at the `service configuration page`_. .. _service configuration page: ../admin/configuration/configuration.html Groups ------ Groups define sets of services that can be manipulated together. Groups are directly linked to rating rules, and not to services or fields. Therefore, if we want to group a set of rules to list them together or delete them, we can create a group and add them to the group, but in the end the resources are going to be charged based on the services, fields and rating rules. Fields ------ Fields define the attributes that are retrieved together with the service collection that can be used to activate a rating rule. PyScripts --------- It is an alternative method of writing rating rule. When writing a PyScript, one will be able to handle the complete processing of the rating. Therefore, there is no need to create services, fields, and groups in CloudKitty. The PyScript logic should take care of all that. Rating rules ------------ Rating rules are the expressions used to create a charge (assign a value to a computing resource consumption). Rating rules can be created with PyScripts or with the use of fields, services and groups with hashmap rating rules. You can define a ``start`` and ``end`` dates to the rating rules (PyScripts and hashmap mappings), which define the period they will be valid and applied in the rating process. If neither ``start`` or ``end`` dates are defined, they will be set as the rule's creation date and ``None`` respectively. A ``None`` ``end`` date means the rating rule will last indefinitely. The ``start`` date cannot be set as a past date, if you really need to do it, you can force it using a ``force`` flag when creating the rule. Once the rule starts running (rule's ``start`` > current date), you will not be able to update the rule anymore, in this case, if you need to change the rule's value, you will need to delete it and create it again with the new value. If we have a hashmap mapping configuration for a service and another hashmap map configuration for a field that belongs to the same service, the user is going to be charged twice, one for service and another for the field that activated a rating rule that is linked to the service. Rating type ----------- Rating type is the expression used to determine a service definition in the collection backend. For instance, one can use the following syntax in the ``metrics.yml`` file. The entry ``dynamic_pollster.compute.services.instance.status`` is the definition for rating types. In the example shown here, there are two rating types being defined, one called ``instance-usage-hours`` and the other called ``instance-operating-system-license``. The rating types are configured in CloudKitty API as services. If they are not configured, they will not be rated by rating rules defined with hashmap. Therefore, they would be collected, and persisted with value (price) as zero. .. code-block:: yaml metrics: dynamic_pollster.compute.services.instance.status: - unit: instance alt_name: instance-usage-hours description: "compute" groupby: - id - display_name - flavor_id - flavor_name - user_id - project_id - revision_start - availability_zone metadata: - image_ref - flavor_vcpus - flavor_ram - operating_system_name - operating_system_distro - operating_system_type - operating_system_version - mssql_version extra_args: aggregation_method: max resource_type: instance use_all_resource_revisions: false - unit: license-hours alt_name: "instance-operating-system-license" description: "license" groupby: - id - display_name - flavor_id - flavor_name - user_id - project_id - revision_start - availability_zone - operating_system_distro - operating_system_name metadata: - image_ref - flavor_vcpus - flavor_ram - operating_system_type - operating_system_version extra_args: aggregation_method: max resource_type: instance use_all_resource_revisions: false