One of two things on my mind that I am constantly faced with during ADDM implementations or Service / Application modeling.
ADDM should support the discovery of software clusters, like clustered http servers, applicaiton servers, database clusters, etc. Some of the most obvious ones that comes to mind are ofcourse the SoftwareInstances created by TKU patterns, that do collect cluster information such as cluster_name etc but doesn´t do anything with the information except storing it as an attribute on the SI.
With CAM initiatives, it is simply impossible to model clustered software instances that are part of an application accurately. Considering that this data later on is consumed by ITSM processes and BPPM it is even more important for prioritizing incidents, planning change etc etc
On Microsoft Windows or Unix Clusters, when a resource cluster switches from cluster node HostA to cluster node HostB, each software instance discovered by ADDM that was running on HostA is removed from ADDM and new Software instances are created (running on HostB).
After synchronization with Remedy CMDB, original SoftwareServer CIs (on HostA) are marked as deleted and new SoftwareServer CIs (on HostB) are created. We are faced with creation and removal of CIs each time a resource cluster switches. As a consequence:
- All manual attributes added to the original Software instance CI are lost (not visible with the new software CI),
- Relationships with people (eg: Support Group having Managed by relationships with the Software are loast,
- Relationships between original software CI and other CIs that were created directly in CMDB are marked as deleted,
- Relationships with Change requests and Incidents are attached to 2 CIs (depending on which host was hosting the software when the Change Request or Incident was raised), rather than one unique
We would like to have a unique Software SI in ADDM corresponding to this software instance running on the cluster, not dependent on the host it is running on.