Practical Configuration Management

  by Martin McEvoy | 11 Apr 2008 | Permalink |

IT management is all about provided IT services to users. For many organisations rapid change, mergers, takeovers and reorganisations have left the services muddled to a point where it is no longer clear what part each Configuration Item (CI) plays in the IT machine.

IT Infrastructure groups are left to manage their CIs without a proper service context and Application Support groups live with vulnerable hosting services.

So where do you start:

1) Impose strict change control – This is known to directly improve service. But applying the highest level of rigor to all CIs across an estate is an expensive approach. Assessing every change properly takes effort and reduces flexibility to a customer’s needs. With limited service information change approvals will always be risky. Excessive change control can stifle progress and demoralise talented staff.
2) Deploy a CMDB – An effective CMDB will provide a service view of the IT estate. But many CMDB projects fail to achieve this because even a proven CMDB product needs data. The data is unique to the organisation and existing data sources from inventories and asset management tools will not achieve the service view.
3) Implement ITIL – Best practice will improve service but ITIL is also dependant on the service view so the problem doesn’t go away.

A practical approach:

1) If we go back to the objective of IT and identify the key services provided to users this is a good starting point. This could take the form of a service catalogue or a list that will later feed one. The result will usually be a bunch of key applications and common services such as desktop, email etc. In my experience the number of key applications is usually much lower than first perceived.
2) If each application and service is modelled using an ADM tool this will provide the magical service view. It will also provide automation to maintain the configuration item records. It is possible to model applications by targeting known hosts then looking at the communications to other hosts and scanning those. A proven iterative process – with a two man team, we recently modelled 174 Applications within 4 weeks using this approach.
3) Once an Application is modelled a solid change management process can be applied to all production CIs within it. Less stringent change control can be applied to non production CIs instances (Development, test etc). Corrective actions can be targeted at CIs within each Application for known configuration issues, end of live issues etc etc. Configurations of Applications can be automatically compared to identify gaps (production and DR environments etc). A good ADM solution will have reports and functionality to support all this.
4) Services and process can be tiered for Critical CIs, non critical CIs and CIs with unknown purpose.
5) Once all Applications are modelled the question is what are all the remaining CIs with unknown purpose doing?? This is a bucket of cost saving opportunities. But expensive change control process don’t need to be applied to these CIs as it is know they do not support critical services. ADM can also auto discover these CIs and provide information to target candidates for decommissioning, virtualisation etc.

This approach quickly gets Configuration Management and other processes shaped around delivery of what is most important. It also isolates the area of waste. Service maps are available for existing tools such as Service Desk systems, change systems, monitoring systems etc.

This will result in reduced outage issues and an improved service offering.

The next step is a CMDB and now the requirements are better known. CI data is ready to support the other processes that the CMDB will tie together. Deploying the CMDB and integrating at a process workflow level is now much more achievable.