I was having a discussion about Pulse the other day and someone brought up the question of whether there was a process, or other way, to migrate specific changes as part of a package. For example, adding specific attributes or element or attribute values. Selecting the changes through the appropriate step in Pulse would not be appropriate in this case.
The thought process is that elements and attribute values for a dimension may be very different between production and development but as part of a deployment there are some changes that were made in development that need deploying. I know I could create a TI that specifically adds the elements and sets the attributes, but is there a generic process that already exists that say references a source file that can be used to specify the changes?
Hey @R.B. I guess there are two (or three) things here
Adding new elements to a dimension - but we assume here you want the dimensions in sync?
Adding a new attribute to a dimension - ditto above
Adding (or updating) data to an existing attribute, essentially a cube data migration
I would suggest the dimension definition source https://forum.cubewise.com/t/mvp-of-the-day-pulse-dimension-monitoring/3780 is the bets way to go for element migration (if you do indeed want ALL new elements), but if you are after say a specific new element (and not others that are still a work in progress) the best way is to create a specific TI to do that for you - and then migrate and execute the TI with Pulse.
The dimension monitoring doesn’t really help in this case as dev has some changes that I want to deploy and some that I don’t. Same goes for the attributes and attribute values. The scenario is that there are several developers working on different projects and not all the changes should be migrated at this point and some of those changes may never be migrated. Trying to select the changes to apply would be error prone in this case, and it’s difficult to test and refine a migration when you do this. I could create a process for each migration that does all the updates, which I’ve done many times, but what I am after is a generic process that loads the changes from a file where I can specify elements to be added, deleted, values to be set and so on. I could write one myself but I expect I’m probably reinventing the wheel here.
The most helpful change in this scenario would be to be able to migrate “dimension header information” (e.g. attribute names and types) independently of dimension structure and attribute values. It is pretty normal that dev/test systems have different metadata versus production which you don’t want to risk pushing to production.
The only current workaround is to include a TI process with no parameters which does actions like creating new attributes. If Pulse could support an easier less labor intensive method it would be good!