Container Survey, June 2016

A note on methodology: This survey was initiated from my Twitter account (@brendandburns) and subsequent retweets. Consequently, the sample from which responses were drawn is likely biased towards early adopters of container technology and Kubernetes. That said, the data in the survey matches up with several other surveys with broader reach (e.g. OpenStack user survey, Spring 2016)

After a brief hiatus, I’ve resumed my monthly survey of the container landscape. One new note, I have created a github repository with the survey questions if you want to add/remove/modify questions, please send me a pull request!

One other thing to note, our sample size this time (229) was more than 2x the sample size last time, so thanks to all who answered and retweeted! Without further ado, what’s the state of the container world?

The last time we sampled in March, we found 80% of folks surveyed were currently using containers, three months later, that number has grown to 84.8%, an additional 12.7% are considering containers. 2.6% indicate they’ll never use containers, down from 5.4% in March.

The predominant use case for containers is still testing (90%) while a majority (60%) are also using containers in production. People are still planning on using containers in production, since when asked what they are planning to do, rather than what they are currently doing, 85% plan to run containers in production.

When asked about the number of containers, the number of large container deployments (>250) shrank to 12.7% from 19.8%, I believe that this is due to the increased sample size reaching a broader cross section of the market rather than people pulling back from large deployments, but we’ll see in the coming months if there is a trend there.

In terms of where these containers run, container services remain popular, with 32.7% of users using some sort of container service. However, growth is mixed, with Google’s Container Engine (19% -> 21%), Microsoft’s ACS (1.4% -> 3.3%) and Docker’s UCP (3.3% -> 4.1%) posting small gains, while Amazon’s ECS lost some ground (15% -> 12%). Container service use seems to be generally exclusive, with only a few people using more than one cloud provider’s container service.

Laptops/workstations 57.7% and physical machines 40.5% remain more popular than any other alternatives. As always, it’s important to remember that these numbers are not exclusive, users are asked to select all categories that apply to their situation. For example, nearly all of the GKE users also deploy containers to their laptops or workstations.

In terms of how containers are run, Kubernetes is still by far the most popular solution for orchestrating containers at 53.4% of all respondents. Shell scripts continue to run strong with 29.5% of users using shell scripts in one form or another.

As always, you can find the raw data here. Please contribute to the survey repository and craft the community survey to go out next month. (and if you find this interesting or useful, please share and retweet). Also if you have particular analysis you’d like to see run on the data, please let me know or file issues in github.