This documentation supports the 20.08 version of BMC CMDB, which is available only to BMC Helix subscribers (SaaS).

To view an earlier version, select the version from the Product version menu.

Creating and testing impact models

You can configure an impact model by creating impact relationships between configuration items (CIs) in CMDB Explorer. You can test this model by running impact simulations in CMDB Impact Simulator to determine how changes to a CI affect other CIs or business services.


CMDB Impact Simulator can predict the impact on CIs by using the impact relationships (both regular relationships with impact and impact-only relationships) that you have configured within BMC CMDB. If you have Truesight Operations Management installed, Impact Simulator can also use the impact relationships configured for Truesight Operations Management.

You can also manually create impact models of services in CMDB Explorer and test the impact on a service when a provider is unavailable. For example, you can test the effect of weighted averaging to a consuming CI in the impact model when one or more provider CIs are unavailable.

The following figure shows a sample impact model created in CMDB Explorer. There are four software server CIs connected to a cluster CI, which is then connected to a technical service CI.

  • The impact relationships in this impact model use the following default settings:
    • ImpactWeight — 100
    • ImpactPropagationModel — DIRECT
  • By default, the ImpactComputationModel of the cluster is set to STANDARD

Sample impact model

If you do not modify the impact weight, propagation model, or cluster type default values, the servers function independently of each other in the impact model. As a result, if a server CI is unavailable, the predicted impact is that the cluster CI and the technical service CI will also be unavailable, even though three other servers are fully functional. 

The following figure shows how a service is impacted if weighted clustering is not implemented and a server is unavailable.

Impact on service if provider is unavailable and no weighted clustering 


You can modify the impact model settings and see how these changes impact a service. For example, you can modify the ImpactComputationModel setting of the cluster to WEIGHTED_CLUSTER. The WEIGHTED_CLUSTER setting uses the ImpactWeight value of each of the providers. If you do not modify the default values of this setting, which is 100, all of these providers are considered equal in the impact model. As a result, when the provider is unavailable, there should is no impact on the technical service.

You can also modify the impact weights and propagation model settings of the relationships. For example, by setting the impact weight of one of the servers to 400 and the propagation model to INCREASING,  that server is now assigned more weight in the computed status than the back-up servers. As a result, when the server is unavailable, there is only a minor impact on the technical service.


Impact model settings

The following table describes the different settings that affect impact simulations.

Setting

Description

ImpactWeight (in impact relationship)

Specifies the weight that this impact relationship has in computing the status of the consumer CI. The default setting is 100. This weight is relative to all the relationships that have an impact on consumer CI. Assigning a weight on an impact relationship enables you to model the effect to the business service if a provider CI is unavailable.

ImpactPropagationModel (in impact relationship)

Specifies the propagation model used for this impact relationship. The following values are available:

  • DIRECT (default) — Propagates status of the provider CI to the consumer CI. For example, if the provider CI status is Unavailable and the ImpactPropagationModel setting is DIRECT, then the consumer CI status is Unavailable.
  • INCREASING — Increases the severity of the status of the provider CI. The status of the consumer CI becomes more critical than that of its provider CI by one level. For example, if the provider CI status is Impacted and the ImpactPropagationModel setting is INCREASING, then the consumer CI status is Unavailable.
  • DECREASING — Decreases the severity of the status of the provider CI. The status of the consumer CI is less critical than that of the provider CI by one level. For example, if the provider CI status is Unavailable and the ImpactPropagationModel setting is DECREASING, then the consumer CI status is Impacted.
  • JUST_WARNING — Propagates only warning information from the provider CI to the consumer CI.

ImpactComputationModel (in cluster)

Specifies the status of the cluster, based on the statuses propagated from provider CIs. The following values are available:

  • STANDARD (default) — Computes the status using the highest propagated value of the incoming relationships. For example, if the provider is Unavailable, the cluster will be Unavailable.
  • CLUSTER — Computes the status where greater than 50% of the providers (the "quorum" percentage) must agree on the same status. If you have four server providers, three of them must have the same status in their impact relationship to the cluster.
  • WEIGHTED_CLUSTER — Computes the status by averaging all statuses of provider CIs and propagating the impact weight to the cluster. The weight determines how much importance (numerically weighted) to give to each provider relationship that impacts a consumer CI. The higher the number, the greater the importance.
    Using the WEIGHTED_CLUSTER setting allows the providers to operate together as a functional whole and increase their availability.
  • LASTALIVE_CLUSTER — Computes the status where if 100% of the providers are unavailable, the cluster is unavailable. The cluster reports warning in the following scenarios:
    -  If one or more providers are unavailable. 
    - If one or more providers report a warning.


To test the settings that affect impact simulations

  1. Edit the propagation model settings of the impact relationship between two CIs (for example, between one of the software server CIs and the cluster CI).
    1. Right-click the relationship and click Edit.
    2. In the BMC_BaseRelationship form, click the Custom tab.
    3. In the ImpactPropagationModel field, type INCREASING.
    4. Click Save.
  2. Run an impact simulation on the server (for example, by setting the simulated status to Very Impaired).
    The INCREASING value in the propagation model increases the severity of the status propagated from the provider CI to the consuming CI. The predicted impact is that the cluster and the technical service will be Unavailable. 

    Impact on service with change in propagation model values 
  3. Edit the cluster.
    1. Right-click the cluster and click Edit to open the BMC_Cluster form.
      If BMC Remedy ITSM is installed, you must use the following direct access URL to open the BMC_Cluster form:
      http://midtierserver/arsys/forms/arserver/BMC.CORE%3ABMC_Cluster
      You then must search for the cluster that you need to edit.
    2. Click the Custom tab.
    3. In the ImpactComputationModel field, enter WEIGHTED_CLUSTER.
    4. Click Save.
    5. Perform the same steps on the technical service.
  4. Run an impact simulation on the server CI (for example, setting the simulated status to Unavailable).
    Based on the new WEIGHTED_CLUSTER value, all of the providers are considered equally in the weighted average. Even when one server is unavailable, the predicted impact is that the cluster and the technical service will experience no impairment, because three other servers are still running. The effect of one server being unavailable has been balanced among the four providers. 

    Impact on service with change only to cluster settings 
  5. Edit the impact weight of the impact relationship between two CIs (for example, between one of the software servers and the cluster).
    1. Right-click the relationship and select Edit.
    2. In the BMC_BaseRelationship form, click the Custom tab.
    3. In the ImpactWeight field, enter 400.
      The ImpactWeight is an integer that is relative to the other providers to the cluster. To change the behavior of the impact model and to provide more weight to the provider in the computed status, you must change the weight relative to the other values. The provider now has four times the weight relative to the other servers.
    4. Click Save.
  6. Run an impact simulation on the server CI (for example, setting the simulated status to Unavailable).

With these new values, one provider is considered four times more important in the weighted average. When that provider is unavailable, the predicted impact is that the cluster and the technical service will experience some kind of impairment.

Impact on service with change to cluster and impact weight settings

When you run the actual simulation in CMDB Impact Simulator, you see that the cluster and the technical service will experience minor impairment: The server's delivery of service is slightly affected.





Was this page helpful? Yes No Submitting... Thank you

Comments