Analysis of 2021 User Survey Feedback

Introduction

This is an analysis of the results from the 2021 User Survey in comparison to the 2019 results. The 2019 User Survey result analysis is being used for comparison as the TC did not perform an analysis of the 2020 User Survey results. As with the previous analysis, the analysis will be performed on a per-question basis with an over-all summary at the end.

How do you upgrade your version of OpenStack?

As with the 2019 User Survey, this is the question that got the greatest number of responses. Since responses to none of the questions are mandatory we need to infer the total number that responded. There were only a couple of respondents that answered other TC questions while leaving this answer blank. As a result, the number of responses, 276, was used as the estimated total number of respondents to the TC’s question.

Response

Number of Users

Percentage of Responses

Upgrade all coordinated releases (once every 6 months)

83

30

Skip/Fast-forward releases

91

33

Not Upgrade

65

23

Deploy Regularly from master branch

13

5

Deploy all intermediary releases and all coordinated releases

20

7

The total number of responses for 2021 was somewhat lower than the 337 responses that were submitted in 2019. There are a couple of positive trends, however, to notice. The first one being that the number of ‘Not Upgrade’ responses decreased significantly. In 2019 35% of respondents indicated they did not upgrade as compared to 23% in 2021. The percentage of users that indicated that they ‘Deployed Regularly from master branch’ also decreased from 9% to 5% while ‘Deploy all intermediary releases and all coordinated releases’ stayed constant at 7%.

The increase in users upgrading their releases appears to be pretty evenly spread, percentage wise, between ‘Upgrade all coordinated releases (once every 6 months)’ and ‘Skip/Fast-forward releases’ increasing from 22% to 30% and 27% to 33% respectively. Unfortunately, we don’t know that the respondents that no longer indicated that they did not upgrade are the same ones that now indicate that they do upgrade. With that said, the decrease does appear to be a positive trend. It will be worth watching these results in subsequent surveys to see if that trend continues.

One other data point worth noting is the fact that we have reached a point where the number of users using the ‘Upgrade all coordinated releases’ and ‘Skip/Fast-forward releases’ options are comparable. The TC sees this as an indication that the changes we have made in the release process to allow doing Skip/Fast-forward upgrades have been worth the effort that was required. As we continue to improve/refine the upgrade options these are data points that we will want to continue to watch.

Once on a given release, do you use stable branches for bug-fix upgrades?

Consistent with the 2019 survey, the number of responses to this question was high, at 265 responses and had a notable total increase from 216 responses in 2019.

Answer

Users

Percentage of Responses

I do not do bug-fix upgrades

46

17

Yes, backporting specific fixes

44

17

Yes, deploying every commit on the stable branch

18

7

Yes, upgrading at various points in time depending on fixes

75

28

Yes, using only official point releases

79

30

The responses in 2021 were very consistent with the 2019 results. The number of users reporting that they do not do bug-fix upgrades or backport specific fixes decreased from 19% and 21% respectively. The number of respondents that indicate that they only deploy official point releases also decreased slightly from 31% to 30%.

While the number of users to indicate that they deploy every commit remained the least reported approach, it did increase from 4% in 2019 to 7% in 2021. Users indicating that they upgrade depending on the fixes available increased from 26% to 28%.

So, overall, the results for this question in the 2021 survey were consistent with those collected in 2019.

To which projects does your organization contribute maintenance resources such as patches for bug and reviews on master or stable branches?

This question had a good increase in responses compared to 2019, with only about 33% of users responding previously compared to 128, or 46%, in 2021. As in 2019, the majority of contributions are made to core projects . There is, however, a new entry in the top five projects: Kolla.

Kolla is an OpenStack community project that provides production ready containers and deployment tools for operating OpenStack clouds. Given the continued growth in popularity of containerizing workloads, the increase in contributors is not particularly surprising and would seem to be a good sign for OpenStack as users are seeking newer and better ways to deploy OpenStack clouds.

All of the projects in the top 5 showed a notable increase in reported participation from the 2019 survey. While Glance was knocked out of the Top 5 projects, it still had a reported increase from 25 respondents indicating activity in the project to 27.

Project

Contributors

Nova

63

Neutron

56

Cinder

40

Keystone

34

Kolla

34

After the first review of these results in the 2019 user survey, the TC wanted to compare the number of respondents who reported they used a project as compared to the number who reported that they contribute to the project. For the 2021 results review we have continued this analysis but have also added the previous percentage of participation for reference. To get this information we considered the number of users who reported that they were using the project in production. Note that we did not count users who indicated that they were just testing a service or who indicated that they had the service installed as part of a Proof of Concept. So the number of users that are using any given service may be notably larger than indicated in the results below. Here are the results of that investigation.

Project

Contributors

Users

% Participation

2019 % Participation

Aodh

3

31

9

25

Barbican

8

70

11

19

Blazar

6

4

150

100

Ceilometer

10

84

12

14

Cinder

40

216

19

9

Cloudkitty

4

12

33

23

Cyborg

3

4

75

100

Designate

13

55

24

25

Glance

27

231

12

8

Heat

13

164

8

8

Horizon

18

224

8

6

Ironic

23

65

35

27

Keystone

34

235

15

9

Kolla

34

69

49

52

Kuryr

5

8

63

71

LOCI

3

7

43

40

Magnum

13

42

31

29

Manila

13

26

50

23

Masakari

2

14

14

17

Mistral

8

23

35

35

Monasca

4

33

12

23

Murano

3

6

50

18

Neutron

56

229

24

15

Nova

63

230

27

15

Octavia

26

89

29

35

OpenStack Client

21

191

11

8

OpenStack Ansible

27

59

46

38

OpenStack Helm

5

14

36

23

Panko

3

7

43

45

QA

2

N/A

N/A

N/A

Rally

6

44

14

16

Sahara

1

2

50

21

Senlin

4

5

80

N/A

Swift

19

102

19

11

Tacker

1

3

33

75

Trove

4

14

29

15

TripleO

11

26

42

16

Watcher

2

6

33

N/A

Zaqar

2

9

22

23

Zun

2

8

25

N/A

A couple of things to note from the results above. The numbers seem to confirm one concern that has been previously raised within the community. There appear to be a number of the smaller projects that are being used and maintained by a small group of OpenStack users. It is also interesting the disparity between the number of users and contributors for the core projects like Cinder, Glance, Horizon and Keystone. It is unclear from these results as to whether the disparity is because these projects have become more stable and don’t require as much development or if there are other reasons for this disparity.

When considering the percentage of participation as compared to the 2019 User Survey results there are a number of things to note. First, there has not been a notable decrease in user participation across projects. Also, it is interesting to note that a majority of projects continued to have consistent numbers as compared to the previous survey. A number of projects even showed an increase in participation. This is true for many of the core projects like Cinder, Keystone, Glance, Ironic, Neutron and Nova. We don’t know if these changes are due to a change in the demographics of the respondents (more user/developers responding vs. just users) or if it is a positive indication in the health of the community. Overall, a key take away would be that project participation has remained largely stable or grown slightly since the last user survey.

How do members of your organization contribute to OpenStack?

As with the other questions in the user survey, this question also saw an increase in responses. 173 users responded for a total of 63% of the total.

Contribution

Users

Percentage of Responses

Bug reports

173

100

Bug fixes on master

85

49

Participate in forum sessions at the summit

59

34

Participate in PTG sessions

44

25

Code review on master

43

25

Particpate in ops meetups

43

25

Documentation improvement

39

23

Backporting bug fixes to stable branches

37

21

Sponsor in-person events

28

16

Code review on stable branches

26

15

Feature design review

22

13

Contribute resources to run CI jobs upstream

8

5

Host third-party CI jobs downstream

4

2

In the 2019 survey the TC noted that the responses are very interesting given that this is a ‘User Survey’. The responses show that OpenStack is unique in the fact that it is a user driven community with many of the users also being contributors. In fact we had more respondents indicating that they contributed to OpenStack than in 2019.

A couple of things of interest to note. First, the number of respondents that indicated they contributed bug reports increased notably from 86% in 2019 to 100% in 2021. More importantly, the number of users that indicated that they contribute bug fixes on master increased from 36% to 49%. This takes this response from 4th most popular way of contributing to 2nd!

Some of the decreased numbers in participation are likely to be due to the pandemic that we all have been dealing with. Reported participation in summit sessions and ops meetups both decreased from 47% to 34% and 39% to 25% respectively. Similarly, sponsorship of in-person events dropped from 20% to 16%. These drops, however, are not as significant as expected given the pandemic, which leads to TC to question these results. For future surveys we may want to look at the wording of the questions. Perhaps users are responding based upon all previous activity, not just the activity since the last user survey.

A decrease in participation that could have been anticipated by many of the community members is in code reviews for both master and stable branches. The number of users indicating participating in these activities decreased by 6% for master code reviews and 5% for stable branch reviews. This is consistent with what has been seen in many of the projects. Code is being submitted but the time required to get patches reviewed and merged has increased. This is also contributed to by a notable decrease in participants hosting third-party and CI job resources.

The rest of the participation responses were relatively consistent with what was reported in the 2019 User Survey.

What prevents you or your organization from contributing more maintenance resources, or makes contributing difficult?

This question is the one that had the most notable increase in responses from 2019. Only 19% of users (69) responded to this question previously as compared to 47% (131) in 2021. This could be taken in a positive light, that more users are thinking about contributing. Unfortunately, the challenges continue to be similar.

Some combination of time, resources and company support accounted for more than half of the responses. Not having the appropriate skills to contribute was also a common theme in the responses. The TC might be able to help with this by raising awareness of the OpenStack Jobs board.

A few responses indicated that they felt that they were not able to contribute because the version of OpenStack that they are on was so far behind the current release. This response seems somewhat surprising as problems found in the older releases may still need to be fixed in the current release. In some cases, it is possible that the code will have changed significantly from the release that is running. With that said, this is a response that probably deserves further discussion by the TC.

The other response that was repeated by at least a few users was communication and timezone barriers. This is a challenge that has been known by the TC and OpenInfra Foundation for some time. Changes have been made to try to alleviate some of this problem but it sounds like it hasn’t worked for everyone.

Other ways users participate:

This was the field that we changed the wording for from the 2019 user survey. There were only two responses submitted: one listing bug reports again and one indicating that they work with vendors to submit requests for enhancement.

So, this field is not getting a lot of input. Worth discussion among the TC as to whether this should be removed or replaced with a different question.

Summary

In summary, the questions being asked by the TC in the User Survey are providing useful information. While it was unfortunate to see the total number of responses decrease, the fact that we had higher participation responding to the TC’s questions is encouraging. Hopefully this trend will continue as the numbers are more meaningful as we have more years of data for comparison.

It is also encouraging to see the numbers moving in the desired direction for a number of the questions that we are asking. The responses to how/when users are upgrading their OpenStack clouds seems to suggest that the efforts of the TC and the community as a whole have had a positive impact on the user experience, or at least their upgrade habits. The overall improvement in user engagement, participating in the community is also encouraging. The data suggests that the OpenStack community continues to be a consistently active one.

All-in-all, the TC is pleased with the continued responses we are getting and would like to continue to collect this data for future analysis.

Additional Resources

The OpenStack Survey Report also provides a graphical overview of the OpenStack Survey results.