“Bad Data.” Those are two words you never want to hear, especially when it applies to your CMDB or IT strategy. And yet studies (and our own experience with our enterprise customers) show that, at any moment up to 40% of the average enterprise’s IT data is ‘bad.’
Data Quality: essential to IT and business
‘Bad data’ is not that easy to define, because it’s generally not just one thing. Think of all the areas that can come into play:
- (Wrong Value) Having a CI Attribute with the wrong value definitely qualifies as ‘Bad Data’, such as what happens when a ‘Server’ CI type that should have a ‘Manufacturer’ attribute of ‘Dell’ is listed as ‘HP’.
- (Manual Entry) The CI attribute is not a validated field, therefore allowing someone to manually type in ‘Deell’.
- (Missing Value) Having no value for that CI attribute for the ‘Server’ could also be considered ‘Bad Data’, such as having an attribute ‘Manufacturer’ with a value of
. - (Wrong Relationships) Defining incorrect CI to CI relationships, such as this CI ‘Server’ is related to this CI ‘Database’.
- (Missing CI) Or the CI ‘Server’ is never even created in the CMDB, so it is missing completely.
And those are just a few examples. So why and how do they happen? While there are multiple scenarios for each of the above examples, let’s just examine one scenario for each:
Wrong Value: A Service Desk person has a drop-down field selected on a CI and tries to scroll down the page, unknowingly changing the attribute value.
Manual Entry: Not enough coffee in the morning, mistyped a key.
Missing Value: There is no one single discovery tool that can pick up all the attributes for a CI. Certain discovery tools are good for discovering certain CI types and certain attributes.
Wrong Relationships: Sometime, certain relationships between CIs are manually mapped or the relationship discovery tools do not provide the entire or correct CI relationship mapping, leaving gaps.
Missing CIs: A new Server is provisioned on a new subnet range or behind a firewall that your discovery tool can’t reach and therefore does not know about.
Data Evolution: the path to data quality
What’s the solution to all these scenarios of ‘Bad Data’? Most companies already have enough tools to resolve their missing data or bad data represented (or misrepresented) in some form from some data repository from some tools across your organization. The key is how to sort it out and combine it.
As an example, if your current discovery tool (i.e. – SCCM) can’t pick up an attribute like ‘Owner’, it might be because that information is maintained in your Active Directory or in ITAM repository. Maybe one tool is known to accurately get information for a specific attribute, but how do you remember, prioritize and automate using that tool to populate that specific attribute over the other tools that also provide information for that attribute (especially if the values are different)?
If you’re able to combine all relevant data pieces (in this case about a specific Server) from all the different tools simultaneously (imagine 10+ sources), and are then able to define logic rules and work out any differences or conflicts between that data, you would arrive at a complete and accurate CMDB. If at that point you still see gaps or a problem, then consider purchasing a tool specific to that gap. But you should ask yourself first, am I just buying tool after tool or is there a more comprehensive, integrated solution?
The Blazent Data Intelligence Platform takes you out of the tool game, where you’re applying different tools to different problems, and gives you complete confidence in both the completeness and accuracy of your CMDB. It’s an automated and easy-to-implement solution that spells the end of Bad Data.
Mike Full is Senior Sales Engineer at Blazent