Research:Cloud Services Annual Survey/2019

Tracked in Phabricator:
Task T238181 resolved

The 2019 survey collected feedback from Toolforge project members and Cloud VPS project administrators on how the services offered can be improved to help their development and maintenance needs.

Participants

The participants of the survey are the members of the Wikimedia Toolforge project and Cloud VPS project administrators. 1857 participants were contacted via email. The emails were harvested from the Cloud Services' LDAP system ("developer accounts") and the Wikitech database. The final list was created by finding the union of the members of the Toolforge project and all users with "administrator" rights on one or more Cloud VPS projects. Users who have opted-out of email contact by other users in their Wikitech preferences or the survey specific opt-out page were excluded. Specific opt-outs for the Annual Survey can be made by users at wikitech:Annual Toolforge Survey/Opt out.

108 email recipients participated in the survey which represents 6% of those who were contacted.

Dates

The 2019 survey was active between 2019-11-25 and 2019-12-13.

Questions and answers

Basic demographics

About how many years have you used Toolforge or Cloud VPS?

Cohort	Percentage
1 Year	20.7%
2-3 Years	24.1%
4-5 Years	27.6%
6+ Years	27.6%

How many tools have you developed on Toolforge or Cloud VPS?

Cohort	Percentage
0-1 Tools	25.3%
2 Tools	27.6%
3-6 Tools	25.3%
7+ Tools	21.8%

How many tools do you actively maintain in Toolforge or Cloud VPS?

Cohort	Percentage
0-1 Maintain	31.4%
2 Maintain	29.1%
3-5 Maintain	23.3%
6+ Maintain	16.3%

About how many hours per week do you spend developing or maintaining tools on Toolforge or Cloud VPS?

Cohort	Percentage
0 Hours	22.1%
1 Hour	21.8%
2-3 Hours	12.6%
4-8 Hours	19.5%
10+ Hours	22.1%

Which Cloud Services platforms do you use?

Derived from answers to the questions "Are you a member of the Toolforge project?" and "Are you an administrator of one of more Cloud VPS projects?"

Cohort	Percentage
Toolforge	61.1%
Cloud VPS	5.6%
Both	24.1%

Do you use Toolforge or Cloud VPS to run or develop tools?

Cohort	Percentage
No	13.9%
Yes	86.1%

Additional demographics

What kinds of tools do you work on?

Tool kind	Percentage
Bots	75.0%
Webapps	72.8%
Apis	30.4%
Gadgets	25.0%
Dataviz	22.8%
Microsites	17.4%
Abuse Ai	1.1%
Ai Systems Deployment	1.1%
Analytics	1.1%
Irc Bot	1.1%
Reports	1.1%
Wikidata Import Scripts	1.1%

Which Wikimedia projects are your tools built to work with?

Project	Percentage
Wikipedia	76.1%
Wikidata	35.9%
Wikimedia Commons	33.7%
All Wikimedia projects	26.1%
Wikisource	9.8%
Wiktionary	8.7%
Wikibooks	7.6%
Wikiquote	6.5%
Wikinews	5.4%
Wikivoyage	5.4%
Wikiversity	4.4%
MediaWiki	2.2%
Wikispecies	2.2%
IRC	1.1%
Wikitech	1.1%

What storage/caching services do you use when running your tools?

Yes/No	Percentage
MySQL/MariaDB	73.8%
Redis	26.2%
SQLite	16.4%
MongoDB	4.9%
PostgreSQL	3.3%
Static files	3.3%
ToolsDB	3.3%
Custom	1.6%
ElasticSearch	1.6%
Filesystem	1.6%
Memcached	1.6%
Wiki Replicas	1.6%

Are you a member of the Toolforge project?

Yes/No	Percentage
Yes	85.2%
No	14.8%

Which programming languages do you use on Toolforge?

Project	Percentage
Python 3	68.1%
PHP	37.4%
Python 2	26.4%
NodeJS	15.4%
Other	13.2%
Perl	11.0%
Javascript	5.5%
Java	4.4%
Mono/.NET/C#	4.4%

When you develop a tool, how much of your work developing code to run on Toolforge is done locally on your machine (as opposed to remotely on Toolforge)?

Answer	Percentage
Almost all of the work	54.4%
Less than half of the work	16.3%
More than half of the work	14.1%
About half of the work	8.7%
I don't know	6.5%

Which source control mechanism do you use to manage your tool's source code?

Answer	Percentage
Git	84.8%
I do not use source control	15.2%

If we could improve one thing in Toolforge in the next year, what should that be?

This a curated summary of the different topics mentioned in the comments. Further commentary in the last section about Freeform comments.

Topic	Mentions in answers
Documentation	13
Support	10
Languages and libraries	8
Performance	8
Storage	7
Deployment	4
Network	4
Docker	3
Kubernetes	3
Logs	3
Monitoring	3
Getting started	2
Local development with Toolforge resources / data	2
Security (subdomains)	2
SSH	2
Other	11

Are you an administrator of one of more Cloud VPS projects?

Answer	Percentage
Yes	29.6%
No	70.4%

About how many years have you used Cloud VPS?

Cohort	Percentage
1 Year	7.7%
2-3 Years	38.5%
4-5 Years	26.9%
6+ Years	26.9%

What do you use Cloud VPS for?

Answer	Percentage
Hosting one or more tools or other public services such as a web app; bot; dashboard; API; etc.	75.0%
Testing and experimenting with software	50.0%
Running one or more MediaWiki instances	37.5%
Running a backing service (database; cache; etc.) for a tool or other public service	34.4%
Conducting data analysis or other large computational tasks	15.6%

Do you rely on NFS for accessing the same files across different servers?

Answer	Percentage
No	61.3%
Yes	38.7%

If we could improve one thing in Cloud VPS in the next year, what should that be?

This a curated summary of the different topics mentioned in the comments. Further commentary in the last section about Freeform comments.

Topic	Mentions in answers
Configuration Management	4
Storage	3
Maintenance	2
Quotas	2
Deployment	1
Documentation	1
Monitoring	1
Network	1
Other	2

Wikimedia Cloud Services is one of many providers of cloud computing resources. What made you decide to choose Wikimedia Cloud Services as opposed to other options?

Answer	Percentage
Access to Wikimedia-specific resources	81.2%
Cost	60.4%
Ease of collaborating with other Wikimedians	56.4%
Ease of use	36.6%
Philosophical or ideological reasons	32.7%
Privacy and security considerations	24.8%

Are you currently employed by the Wikimedia Foundation as an employee, contractor, vendor, or intern?

Answer	Percentage
No	86.8%
Yes	13.2%

Qualitative questions

A series of qualitative questions were asked in order to gauge general satisfaction with Toolforge and Cloud VPS.

When I look for documentation on Wikitech about Toolforge or Cloud VPS I usually find what I was looking for.

The documentation I find on Wikitech about Toolforge or Cloud VPS usually answers my question.

The documentation I find on Wikitech about Toolforge or Cloud VPS is easy to read and understand.

If you would like to share more about your experience finding, reading, or maintaining technical documentation on Wikitech, please do so here.

This a curated summary of the different topics mentioned in the comments. Further commentary in the last section about Freeform comments.

Topic	Mentioned in comments
Out of date	6
Getting started	3
Hard to find	3
Praise	3
Lacking (DB Schemas)	2
Scattered	2
Support	2
Unclear organization	2

Services provided by Wikimedia Cloud Services, including Toolforge and Cloud VPS, have high uptime

It is easy to have code run on Toolforge or Cloud VPS

I feel that I am supported by the Cloud Services team when I contact them via cloud@lists.wikimedia.org, the #wikimedia-cloud IRC channel, or Phabricator.

I receive useful information via the “cloud-announce” and/or “cloud” mailing lists.

Free form responses

The survey included several free form response sections. Survey participants were told that we would only publicly share their responses or survey results in aggregate or anonymized form.

If we could improve one thing in Toolforge in the next year, what should that be?

Responses fell mainly under 5 main categories:

Documentation (14 comments)

The most mentioned topic was having more and better docs for getting started and introductory materials. Closely following and related was having more documentation about how to use a bigger variety of languages and libraries. And finally, some comments about having more video material (tutorials, talks and presentations).

Support & Community (12 comments)

These comments mentioned more support for beginners via documentation and tutorials, having places for discussion, resolving problems and knowledge sharing, better direct support from the team.

Platform (40 comments)

Within this theme, there is a wide variety in the comments.

Top mentions are better support for more languages, software and libraries, both having the latest versions and the ability to use other languages and libraries (and having more varied documentation with this regard). And also multiple mentions of improving performance of SSH sessions, file system and WikiReplicas.

Related but separately, there were multiple comments mentioning improving storage, both capacity and performance, of various DBs and the file systems (NFS).

There were also some comments related to workflows, that will be summarized in the next section, related to platform wanting to have other deployment mechanisms related to Docker images and Kubernetes.

Other comments not included in previous mentions talked about improving security, particularly around sharing subdomain.

The rest of the comments cannot be anonymously summarized because of their small numbers.

Workflow (13 comments)

In the workflow category, the most mentioned topic was deployment, having easier and more flexible options for deployment (git or docker based for example).

Following is observability, people mention both logs and monitoring, easier access and setup of logs and error reporting, and ways to monitor their projects.

Lastly it was mentioned wanting to have a way to develop software locally having access to Toolforge's data.

If we could improve one thing in Cloud VPS in the next year, what should that be?

We have less responses here so the summary will be a bit broad.

There is one clear ask mentioned in many comments, on the improving the configuration management side (better puppet support, more flexibility, more options).

Another topic mentioned multiple times as well is improving storage (faster, more space, easier maintenance when full).

The rest of the responses fell on a variety of topics: easier maintenance, quota related, deployment, documentation, monitoring, network, etc.

If you would like to share more about your experience finding, reading, or maintaining technical documentation on Wikitech, please do so here.

The main complaint was about documentation being out of date, followed by being hard to find, and related, having a hard time getting started.

There were also comments praising the quality of the documentation (even if old and hard to find), which are greatly appreciated. Thank you.

Other topics mentioned including having better documentation for the DB schemas, the docs being too scattered or having unclear organization.

If you have any other comments, please share them with us here.

Here there was a bunch of praise for the improvements and the services provided, which is greatly appreciated. We want to continue improving and doing a better job where we can. Thank you!

There were also some comments related to confusion with some of the questions, and some other comments that can't be shared anonymized.

Thanks to everyone that took the time and participated in the 2019 survey. See you soon in the 2020 one!