best practices for workflow promotion from dev -> prod?

[stephenrutledge]

Hi all,

We are setting up a new Designer cloud account and looking into the best way to support our development lifecycle:

(1) developers create workflow in dev workspace against dev databases

(2) once tested and ready for production, workflows are moved to prod workspace where they are scheduled to run against prod databases.

Current theory how best to do this is:

* create dev alteryx workspace, configure with connections to dev databases

* create prod alteryx workspace, configure with connections to prod databases

* connections in dev and prod databases have same names, but point to dev and prod respectively:
ex:

connection xdb points to database xdb_dev in dev workspace

connection xdb points to database xdb_prd in prod workspace

Then, to promote a flow from dev to prod:

(1) developers create workflows in dev workspace

(2) when ready to promote to production, developers export from dev workspace and import to prod workspace.

Since connections are the same name between dev and prod workspaces, the imported flows will run in prod against the production databases and run in dev workspace against the dev databases.

QUESTIONS: 

Is this a working approach? If not, how do other shops setup a development lifecycle? Are there any best practices docs about various ways to manage Alteryx Cloud designer workflows in a development lifecycle? 

[davidskaife]

Hi @stephenrutledge 

This should provide some guidance on an approach to this - https://help.alteryx.com/aac/en/developer-help/sdlc-asset-promotion-guide.html#sdlc-asset-promotion-... 

[KGT]

As per the link in the comment from @davidskaife , there is one thing to look at still, and that's whether the developers do the push themselves or the admin handles it. If the admin handles it, then the developer would mark it as "Production Ready" or similar, and then Admin would have a process that looks for all Production Ready workflows and upload them to Production via API.

It requires making sure everything is set up better and so may have some upfront time cost, but it also allows you to put extra checks in the process.

Some that I've seen are:

• tool counts,
• notifications on Python/R tools that are in the workflow,
• Checks on Input/Output tools to see dependencies, etc