Developer Satisfaction Survey/2023/Deployment
📖 Developer Satisfaction 2023
The developer satisfaction survey is an annual survey of the Wikimedia developer community.
🚂 Deployment
edit🔗 Deployment use in the past year
We asked survey takers, “In the past year, have you used the Wikimedia deployment train or backport windows?”
- About half (49%) of respondents said they used neither
- About a third (32%) said they used both
- 11% said they used Wikimedia deployment train only
- 8% said they used backport windows only
🔗 Deployment use in the past year (by role)
The groups whose majority used both Wikimedia deployment train and backport windows in the past year were:
- MediaWiki Core developers
- MediaWiki Extension developer
The groups whose majority used Wikimedia deployment train and/or backport windows in the past year were:
- MediaWiki Core developers
- MediaWiki Extension developer
- SRE/Infrastructure engineers
- Tool developers
The groups whose majority did not use Wikimedia deployment train or backport windows in the past year were:
- Data engineers
- Gadget developers
- Service/API developers
- Test engineers
🔗 Deployment train frequency effect on work
💬 There should be a general rule to allow deployers to deploy patches they seem fit any time they want as long as they are not stepping on another deployer's toes and it's not a no-deploy day.
We asked survey takers how often deployment train frequency affected their work during the past year.
- Almost half (47%) of respondents said their work was never or very infrequently affected by deployment train frequency
- 24% said their work was occasionally affected
- 22% said their work was somewhat or very frequently affected
- 7% were unsure
🔗 Deployment train frequency effect on work (by role)
The groups whose work was least affected by deployment train frequency (i.e., whose majority said their work was never or very infrequently affected) were:
- Gadget developers
- Tool developers
The groups whose work was most affected by development train frequency (i.e., whose majority said their work was occasionally, somewhat frequently, or very frequently affected) were:
- SRE infrastructure engineers
- Data engineers
🔗 Backport window frequency effect on work
💬 "scap backport" is brilliant, thank you to the team for making this!
We asked survey takers how often backport window frequency affected their work during the past year.
- The majority (52%) of respondents said their work was never or very infrequently affected by backport windows frequency
- 23% said their work was occasionally affected
- 15% were unsure
- 10% said their work was somewhat or very frequently affected
🔗 Backport window frequency effect on work (by role)
The groups who work was least affected by backport window frequency (i.e., whose majority said their work was never or very infrequently affected) were:
- SRE/Infrastructure engineers
- MediaWiki Extension developers
- MediaWiki Core developers
- Gadget developers
All other groups had varying degrees of affect and higher rates of uncertainty. Data engineers had the highest percentage of respondents who said their work was very frequently affected (20%) and the highest percentage of respondents who were unsure (40%).
🔗 Kubernetes use in the past year
We asked survey takers, “In the past year, have you deployed software to Wikimedia’s production Kubernetes infrastructure?”
- The majority (69%) of respondents said that they did not deploy software to Wikimedia Kubernetes in the past year
- 21% said that they did
- 10% were unsure
🔗 Kubernetes use in the past year (by role)
The two groups with more respondents indicating they deployed software to Wikimedia Kubernetes than not were:
- Data engineers
- SRE/Infrastructure engineers
All other groups had a majority indicating they did not deploy software to Wikimedia Kubernetes in the past year.
🔗 Kubernetes satisfaction
We asked survey takers, “How satisfied are you with software deployment using Wikimedia’s production Kubernetes infrastructure?”
💬 When it works according to plan, it's great. When I get errors, I usually need to ask [...] Thankfully, there's usually someone around to help quickly.
Of the survey takers who had used Wikimedia’s production Kubernetes infrastructure during the past year,
- The majority (54%) of respondents said they were satisfied with using Wikimedia Kubernetes
- 18% were unsure
- 15% were neither satisfied nor dissatisfied
- 13% were dissatisfied
🔗 Kubernetes ease of deployment
💬 Once your project is setup deployments are easy, setting up a project still requires to juggle with many steps
We asked survey takers, “How much would you agree with the following statement: Deployments using Wikimedia’s Kubernetes infrastructure are easy and pain-free?”
Of the survey takers who had used Wikimedia’s production Kubernetes infrastructure during the past year,
- 31% agreed that deployments were easy and pain-free
- 26% neither agreed nor disagreed
- 23% disagreed
- 20% were unsure
🔗 Puzzles in this section
A few respondents raised challenges or problems with this section.
💬 I feel like you were missing a question like “Deployments using Wikimedia’s [non-Kubernetes] infrastructure are easy and pain-free?”
💬 I chose "I'm not sure" on whether deployment train frequency affected my work, because I didn't understand the question. Is there an implied "negatively?"
💬 In this section, it's not clear what "use" means. Did I make use of the deployment [t]rain if I was waiting for a patch to get deployed? The train frequency affected my work sever[e]ly, but I don't feel like I have "used" the train. Perhaps the idea was to ask whether I [depended] on it?