Populating the CMS
The CMS is only useful if it is accurate. In fact, the most dangerous situation with a CMS is when the data is wrong, which is why the concept of federation, as described in Chapter 4, is so critical to your design. Decisions are based on the CMS, and when the data is wrong, the decisions are wrong. Instead of the CMS being an enabler for improvement, it can actually cause further deterioration when the data is suspect. CMS population, therefore, is among the most important of all facets of Configuration Management. How you populate your CMS and keep it accurate will directly affect the success or failure of the CMS and—as we pointed out in Chapter 1—the entire IT operation itself.
The CMS is populated in two ways: manually and automatically. The majority of CMDB population so far has been manual. This is one main reason most CMDB initiatives have suffered or failed altogether. Manual population is risky because it is too difficult to maintain its accuracy. By the time population is finished, the contents are already partially obsolete.
To optimize CMDB accuracy, you want to automate as much of the population as possible. We call this automated discovery, or auto-discovery. For our purposes, we refer to discovery as the automated population mechanism.
Discovery is a wonderful innovation for the CMDB, but alas, many CMDB elements cannot be discovered. Therefore, you will inevitably have a mix of both population modes. The following figures show how this mix works. Figure 2.9 is a simple diagram showing a collection of CIs in a CMS. It is merely illustrative of the point, not an actual CMS.
Figure 2.9 CIs in a CMS
You need to identify what can be discovered and what must be manually populated. By segmenting the CMS in this way, you can set forth with your plans to build automation technologies and operational tasks to build and maintain the CMS. Figure 2.10 shows how the CMS is divided between the two. Discovery usually gives you the core elements, whereas manual methods are used to supplement this core.
Figure 2.10 Two modes of populating the CMS
Both modes can be further broken down into the relevant CMS domains, which are most effectively aligned with the MDRs. Figure 2.11 shows this additional breakdown. Note how many domains will have both discovered and manual components.
Figure 2.11 A further breakdown of CMS population
Domains like the network lend themselves well to discovery because the common instrumentation (for example, SNMP) is pervasive and full of useful data. Others, such as business services, are more heavily manual because of a lack of instrumentation. As we explained earlier, even these domains are improving for discovery. As new technologies emerge to enable discovery, you should capitalize on them to continue building more accuracy into the CMS.