Ways to validate data in the CMDB without Discovery

[Kapil23]

Does anyone have any ideas to QA data in the CMDB when Discovery is NOT turned on?

[HugoFirst]

Since there's a lot of different data in the CMDB, there are a multitude of ways to verify/validate it.

Here's what I do:

We have a focus on System names.

I export a list of servers from the CMDB each day. I then developed a mini-web app that runs under an apache that gives me the following capability:

1. I can paste in a list of server names or IP addresses. ( one per line ).

2. The script will compare the list of names/IP addresses with the exported list of servers from the CMDB.

      If there is no   matching server in the CMDB, the script attempts to PING the server and checks for a DNS name.

3. My output provides the following:

    - a list of the servers found in the CMDB.

    - a list of servers NOT found in the CMDB, not responding to PING and which do NOT have a DNS Name.

    - a list of servers NOT found in the CMDB, but respond to PING and have a DNS Name.

I focus mainly on the third list, if a server in not in the CMDB, but in the network, there's a very good change that it should.

Now your probably asking this question, "Where do you get the list of servers/IPs, to paste in to the web app?".

The answer is ANYWHERE YOU FIND A LIST OF SERVERS.   I get emails from a variety of people informing me of maintenance on servers, or a list of servers being monitored, etc.   I can easily paste in the list from any of those sources and in a few minutes get the result.

[akash_mehta]

Kapil:   So, when Discovery is not in place, the question becomes what other sources of truth are out there?

1)   SCCM, Altiris, HP UD?   Are any of these tools in place in your environment that you can utilize to bring in Configuration Data?

2)   Do you have other tools that manage certain pieces/classes of data that you can at least export from and import to ServiceNow?

3)   Data Certification - using this will help with identifying groupings of data - and assign to users/groups that are responsible for those CIs.

Data Certification

When a discovery source is not in play - make sure you keep it simple.   It is very difficult to manual maintain a CMDB that is very complex.

-Akash

[DaveHertel]

This "QA data in CMDB" question is old, so I'm sharing a feature available in the latest platform versions (London, Kingston, etc.).  If CMDB & CI data quality is important to your company, take a look at using the CMDB Dashboard capabilities.  This flexible dashboard empowers you with visibility into the health of the CMDB (perhaps some would consider this "QA" as mentioned in the original question).   The CMDB Health Dashboard in London rolls up quality metrics based on flexible criteria you define -- from 3 Metric arenas:

• COMPLETENESS
  Determined by CI data field completeness based on 2 criteria: Required fields and Recommend fields.
  Required = mandatory at the database level, while Recommended are desirable fields for the business, but not absolutely needed
• COMPLIANCE
  Determined by the desired state, which is defined by Certification Filters/templates (comparing actual state values vs. desired)
• CORRECTNESS
  Determined by 3 sub-areas you also set to suit your needs:  Duplicate CIs;  Orphan CI relationships;  Staleness/Old CI's
  
  Example of CMDB Dashboard.  This screenshot is focused on entire CMDB, but it's also possible to drill into any class in the CMDB hierarchy to get a more granular perspective of CI health for any CI Type

Hope this helps?    Consider the CMDB Dashboard to establish baselines and evaluate the "quality" of a CMDB implementation.

[Hemashekar]

Hi Dave,

 I am working extensively on CMDB Health Dashboard to enhance my org CMDB.

I have follow up question 

1. On the Required Stat-- total number "6492" , Is this total number of records satisfying the Health Inclusion Rule filter applied for the CI Class for example : cmdb_ci_hardware table. Condition : Class is one of Windows or Linux

2. What happens to this total number if inclusion rule is applied to child class lets say for "cmdb_ci_linux_server" condition as "Status=Operational". wihch as records of 2000

Will this number reflect aggregate of both inclusion rule or only child class. How is this determined.

Thanks in advance

Shekar