We our at the moment implementing SLM in SD 5.x. Some points that can help in this process are (my impression):
- A good point to start is the SLM Guide and certainly SLM senarios chapters. - If you really want to define your own metrics you need to use the Open Metric Adapator, this is a little tricky to configure. Doesn't always work like you suppose it to work but it makes SLM very flexible so for me a must have. - Some Default SD metrics are intersting but not usuable for defining levels on e.g. "Percentage of calls solved with deadline" = 0 if no call is open anymore. - For the SLA,SLO, Metrics, Service Levels each have its own item with own status ans management. All should be set active and are in the actual start - finisch interval to be captured or included in reporting.
I hope this already give you some idea on how to implement SLM.
Hi Iâ m in the process of implementing SLM and Iâ m not quite sure if Iâ m gonna like the result :)
My problem is Iâ m taking an â OLA focusedâ approach to SLM â in general my concept was:
-make the technical teams (Network, DBA etc) put their stuff in CMDB -define a set of request each team can be assigned (kind of low level service catalogue) -divide the stuff in the CMDB according to criticality (so production goes to one group, development/test to the other) -Match the service groups (e.g production databases, test databases) with service levels -Create SLAs (being de fact OLAs) that would be tied to the specific IT teams, so I could sell my teams services to the others
Now the process would be like that: A person from inside the IT would open a service call and choose a CI. The CI would be associated with the service/service level and so the appropriate impact/priority/deadline mapping would be applied. I would have the opportunity to watch the SLA status and know how the teams are doing.
What makes me feel bad about SLM in SD ?
- I canâ t offer the same service to the same organization under different service levels. In my case the metrics like completion time for a given type of request is tied to the level I set for the CI â I want requests to be completed faster on production boxes than on test/development. That leads me to creating same service as many times as the number of service levels I want to offer - The request we got cataloged so far are of very different type (from login creation to a complex hardware operations) and so the amount of time needed to fulfill a request can vary. If I want to accommodate for that I would need to create a service for type of request and than multiply by the number of CI tied service levels. To make things even worse how do I link those service with CI ? I can only define one â managed byâ service for CI and I use that one service to link the availability metrics via Metric Adaptors. I could use UI rules but that would be a night mare maintaining
I think I will go for the approach when I do have all that services defined (say I got 15 request types and 2 service levels for prod and test â that would lead me to 30 different services) and I would use UI rules to set the appropriate service. The user would have to choose a service call category and the rule would check for the type of CI and fill in the service and service level.
One thing is certain that whole SLM thing is complicated :)