Questions I can ask the lead DORA report investigator in an interview?
Posted by shivamchhuneja@reddit | ExperiencedDevs | View on Reddit | 12 comments
Hey everyone, we are soon interviewing the lead investigator from DORA team at Google Cloud and are trying to pose hard questions: just get away from the typical yearly dora report discussion.
Wanted to see if you have anything to share around what we could ask? (I've got my list but thought it might be a good idea to take your questions/opinion on this too)
Specialist-Menu9892@reddit
DORA metrics are usually being calculated based on Pull Requests. How does one calculate them when following the practices of XP - working in pairs and directly merging to main?
jayantbhawal@reddit
My 2 cents, not to be taken as universal fact:
I think there's a minimum team size when something like DORA even begins to be useful.
If there's only two people in a team, I imagine a lot of "formal" process are let go for the sake of getting things shipped faster.
Reviewing code async might be replaced by getting on a quick call to get some feedback, or sitting together to get some things addressed.
But for other larger teams where productivity is actually a priority, they're probably already creating pull requests and doing proper code reviews involving one or more reviewers.
shivamchhuneja@reddit (OP)
Interesting question, PR is a deployment, so even if its deployed directly to main, it can count as such - might need tweaking though - an open source solution should help here easily. Check Middleware Open Source on gh or google search.
I'm sure there are other ways too that someone can expand on.
ashultz@reddit
The really hard question is do all these measurements and reports actually turn into meaningful improvements.
You can follow up by asking which measurements have started to be gamed; how and why?
shivamchhuneja@reddit (OP)
Great, thanks! Btw deployment frequency gets easily gamed as well - single PRs get broken down into micro PRs in and around performance reviews to boost things
Sea_Neighborhood1412@reddit
Deployment frequency is actually included as a metric BECAUSE it will incentivize people to reduce batch size. Decreased batch size is the goal, deployment frequency is the proxy metric that drives the outcome.
ashultz@reddit
but like most DORA metrics it's amazingly good for your og to go from 1 to 100 and probably dysfunctional to go from 100 to 120
shivamchhuneja@reddit (OP)
Another important point/question could be: dora metrics aren't everything, so what else is important for teams after a point - although the reports cover a few different items but in terms of DevOps KPIs on a team level, what else works :)
Sea_Neighborhood1412@reddit
One would hope someone lends an eye to the difference between the positive practice of “deploying small slices of functionality daily or more” and “triggering the pipeline by writing a script that commits one line comments to the code base on a 2 minute loop” 😅
Ciff_@reddit
You get what you measure
b1e@reddit
Hate to break it to you but DORA was considered kind of a joke within Google during my time there and it’s still pretty low ROI for everyone.
I’ll second the other commenter that suggests asking for real world examples of actionable improvements. Often the metrics etc. are so high level or detached from the nature of the work itself that it doesn’t lead to any meaningful improvement
shivamchhuneja@reddit (OP)
Valid point and a good question for sure🫡🫶