How to record value domains for Object representations

Hi all, we have come across the need to define some data elements that have representations of binary large objects and character large objects (BLOBs and CLOBs). For example, a data item that is represented as a Character Large Object, in XML format and size 2MB.

We are trying to determine the best way to name and define the value domains for these data elements. We have a few ideas, but wanted to check with the community for what approaches may already have been adopted?



Hey @skew - good question! It took us a few days to come up with some ideas on how to resolve this.

Our recommendation here is too look at the “BLOB” and “CLOB” as a Data type, as this will identify across the registry where these large objects are used.

These can then be attached to Value Domains that better describe what the contents of these objects are in specific use cases.

For example, below are a “JPEG file” and “BLOB” metadata examples that separate the two for reusae and interpretation.

As a few basic recommendations, you would see things like:

  • Expected Value Domains for the Data Type of CLOB could be text files up to 2GB eg text files, PDF documents, word documents etc.
  • Expected Value Domains for the Data Type of BLOB could be non text files up to 2GB videos, images, gifs, and audio files.

Hope that helps!


Makes sense Sam.
So I’m guessing create a new Data Type called CLOB.
Then your newly created value domain has a Data Type of CLOB, a format of XML … and a name like “CLOB XML”. Data Elements which store large chunks of XML ought to be linked to the CLOB XML value domain.

When dealing with a similar CLOB that for example stores JSON instead of XML, you have a different value domain called something like “CLOB JSON” (which also has a Data Type of CLOB, but the format is “JSON” not “XML”).

In these cases, I’m not sure what the representation class for these value domains would be. We have a rep class called “Binary Object”, so perhaps a “Character Object” could cover off all these large character fields of various formats?


Yup! The idea being that the data type will tell you where all of these large character objects are stored, and the value domains hold information about what sort of data it contains (eg. JSON, XML, etc…)

1 Like