[scarface74]

And when you get ready to rollback, can you just revert all of your code and deploy? Can you just do a “git branch” and work on your own isolated code in your dev account?” How well does merging work when you do a pull request and have to merge your code and make sure your stored procedures are up to date?

[pjmlp]

Yes, because there is such thing as CI/CD pipelines, and test environments, for database code as well.

[scarface74]

If there are ten developers with 10 feature branches are they going to be running 10 instances of Oracle with their own “branches” of the stored procedures?

[pjmlp]

Yes,install the RDMS into their computer just like any other development tool.

Yes, learn to use multiple schemas and instances, just like using multiple deployments into Apache, Tomcat, IIS,...

Or even better, join the hype, and use containers with volumes configured for each feature branch.

[scarface74]

And then copy “enough” data on their computers to have a realistic test and make sure you have some type of script so sensitive data isn’t on their laptops and ensure schema changes are constantly propagated to each developers workstation. This isn’t even mentioning if you have a commercial database like Oracle where you have to deal with licenses.

And how do you merge changes and do rollbacks?

Not to mention the whole

GetCustomer_1

GetCustomer_2

GetCustomer_3

Where code is copy and pasted and modified slightly.

[pjmlp]

No difference to any other kind of programming workflow, same kind of care is required and it is DevOps job to make sure it is taken care of.

Licenses are peanuts in enterprise projects.

[scarface74]

It’s much different. As someone who is a long time developer who also deals with the “DevOps” side of the world. I never design architectures of the former (development) without thinking about the latter from day one.

You can’t just wave your hands and throw it over to the “DevOps” folks. Doing so is kind of just the opposite of what “DevOps” is suppose to be.