Overview of normalization modes
You can configure the Normalization Engine in different modes to normalize CIs. The options available are:
Inline normalization is a real time method of normalization whereby CIs are normalized as soon as they are created or modified, and before they are saved in BMC CMDB.
Inline mode is used mainly for integrations when a data source is writing to CMDB, and during the data population process.
In this mode, the CIs are normalized after they are saved in BMC CMDB based on a schedule for a dataset. You should select the batch mode when normalizing a large amount of data, and schedule it to run outside of heavy use hours to minimize the impact on users.
Batch normalization normalizes product attributes, including categories, regardless of the normalization type setting for CIs with a
NormalizationStatus of Not Normalized and Not Approved.
In this mode, CIs are normalized after they are saved in BMC CMDB based on changes to the CIs, not to the dataset. When CIs are added or changed, BMC CMDB notifies the Normalization Engine which then checks and normalizes the modified CIs. You should use the continuous mode for dataset updates.
Difference between inline and continuous normalization
The following figure shows the difference between inline and continuous normalization.
In this example, Calbro uses inline normalization on CALBRO.APPS because it is not frequently updated. Normalizing CIs one at a time would have minimal performance impact on users. Calbro uses continuous normalization on the CALBRO.DISC dataset for specific reasons. First, Calbro completed a bulk normalization with a batch job. Second, because the discovery tool typically adds or changes few CIs in the dataset, Calbro sets this to continuous mode. Calbro staff also sets normalization to start when 10 CIs are changed or created or when five minutes have elapsed since the previous normalization.
What happens if normalization job in batch mode is interrupted?
If a normalization job in batch mode is interrupted, the Normalization Engine tracks the classes that were normalized before the interruption and resumes with the class that was being normalized. If a class that was normalized before the interruption has a changed or new CI, the Normalization Engine does not scan for changes. Instead, the next job normalizes the changes.
The following figure shows an example of an interrupted normalization job. Classes 1A and 2B were completed before the interruption. Normalization resumes with class 3C, and the normalization job is not aware of new or modified CIs in the completed classes.