Highlights of the 2019 CNCF China Cloud Native Survey Report

10 min readDec 1, 2020

Step up the digitalization of your business with Alibaba Cloud 2020 Double 11 Big Sale! Get new user coupons and explore over 16 free trials, 30+ bestselling products, and 6+ solutions for all your needs!

By Alibaba Cloud Serverless

72% of Respondents in China Use Kubernetes in Production

For a better understanding of the use of open source and cloud native technologies, the Cloud Native Computing Foundation (CNCF) conducts community surveys regularly. This is the third China Cloud Native Survey conducted in Mandarin Chinese, aiming at gaining a deeper understanding of the methods of cloud native technology adoption in China. This survey also aims at understanding how cloud native technologies empower developers and make changes in a large and growing community. This report is based on the previous two Cloud Native Survey Reports released in March 2018 and November 2018.

Key Points of the China Cloud Native Survey

• 49% of respondents use containers in production, and other 32% are planning to do so. This is a significant increase compared to the survey results showed in November 2018, when only 20% of respondents used containers in production.

• 72% of respondents use Kubernetes in production, which is higher than 40% of that in November 2018.

• The usage of public clouds has dropped to 36% at present from 51% in November 2018, while 39% of respondents choose to use hybrid clouds.

• The number of CNCF projects has increased exponentially. CNCF now has four projects born and widely used in China. They are Dragonfly and KubeEdge in the incubation phase, and newly graduated Harbor and TiKV.

The 2019 China Cloud Native Survey included 300 respondents. 97% of them were from Asia, mainly from China.

Container Usage

It is well known that containers have changed the cloud-based basic infrastructure. However, in the past year, the use of containers in production has become the norm. According to the 2019 Global Cloud Native Survey Report released earlier this year, 84% of respondents use containers in production. Therefore, containers are used across the world.

The survey in China shows that although China’s container usage lags behind the world, its momentum is increasing. According to the survey, nearly half (49%) of respondents use containers in production, leaping from 32% in March 2018 and 20% in November 2018.

The number of respondents planning to use containers in production decreases to 32% now, while it is 57% in March and 40% in November 2018. This means that many organizations have put the container plan into practice instead of in the planning stage. There is still room for a continuous growth.

As container application proportion in production increases, there are fewer containers used in the test environment. About 28% of Chinese respondents currently use containers in the test environment. The result shows a slight increase from 24% in March 2018, but a decrease from 42% in the November 2018.

Although containers have brought amazing advantages, they also brought certain challenges. Changes have taken place over time, but the challenge of complexity has remained unchanged. In the survey, 53% of respondents list complexity as the most significant challenge. In contrast, in the survey in March 2018, 44% of respondents believed so, accounting for the highest proportion. In November 2018, that proportion was 28%, accounting for the third place of all challenges.

In terms of challenges, security ranks second with 39% of proportion. Security is given top priority for the first time. Challenges of lacking of training and network rank third, accounting for 36%. 35% of respondents regard reliability and monitoring as deployment challenges.

Growth of Kubernetes

As a general platform for container orchestration, Kubernetes is emerging in the industry and has rapidly increased in adoption in the CNCF community in China. 72% of respondents say they use Kubernetes in production, showing a significant increase from 40% of that in November 2018.

As a result, the number of respondents evaluating Kubernetes decreases from 42% to 17%.

The survey report also shows that the growth of Kubernetes production clusters in two deployment number section. Most respondents in China use less than 10 clusters, but the number of respondents using over 50 clusters has increased. This may be a result of the increase in the number of new respondents using containers in production.

36% of respondents have 2 to 5 clusters, higher than 25% of that in November 2018. Besides, almost 50% of respondents use 1 to 5 clusters, and 70% use 1 to 10 clusters. Only nearly 13% of respondents have more than 50 clusters in production, compared with 5% of that in November 2018.

Packaging

Helm is the most popular method for packaging Kubernetes applications, with 54% of respondents choosing the Helm.

Entry

NGINX (54%) is the most popular Kubernetes entry supplier, followed by HAProxy (18%), F5 (16%), and Envoy (15%).

Kubernetes Applications Detachment

Managing objects in a cluster is challenging, but namespaces could facilitate management through filtering and controlling by group. 71% of respondents use namespaces to detach Kubernetes applications. From respondents with several teams using Kubernetes, 68% of them use namespaces.

Monitoring, Logging and Tracing

For those users who use solutions of monitoring, logging and tracing, it is more common for local operations to be hosted through remote servers. 46% of respondents use local monitoring tools, while 20% run local operations through remote server hosting. Fewer respondents adopt solutions of logging and tracking, but 26% of respondents run tracing locally and 20% through remote servers. In addition, 21% of enterprises run tracing tools internally, while other 21% through remote servers.

Code

With the support of continuous integration (CI) and continuous delivery (CD), powerful functions of cloud and containers have jointly promoted the speed of development and deployment in China. The survey quantifies the development speed by the code checking-in frequency of developers. 35% of respondents check in code for multiple times a day, 43% check in for a few times a week, and 16% check in for a few times a month.

The release frequency of most respondents (43%) is once a week, while only 21% perform releasing once a month, and 18% once a day. 12% of respondents release on a specific schedule.

CI and CD

Many people think that the foundation of successful CI and CD is process automation. However, the survey in China shows that there are relatively few pure automation environments. Only 21% of respondents adopt automated release cycles, while 31% rely on manual release. The hybrid mode is most popular, accounting for 46%.

CI and CD is a technology for flexible delivery and lifecycle management of cloud native systems. Jenkins is the most popular CI and CD tool in Chinese community, accounting for 53% in community and 40% in GitLab.

Cloud and On-premises

The use of clouds is growing. However, the survey shows great changes of public, private and hybrid clouds. Public cloud usage appears to peak at 51% in the November 2018, dropping to 36% this year. Private cloud usage remains nearly unchanged and occupies 43% in November 2018. It’s worth mentioning that hybrid cloud is the new choice this year, accounting for 39%.

Cloud Native Project

CNCF manages a large number of open source projects. These projects are crucial to the development, deployment and lifecycle management of cloud native. CNCF projects are growing exponentially in China. For example, 57% of respondents use Prometheus monitoring and alarm systems, greatly increased from 16% in March 2018. 35% of respondents now use CoreDNS, while only 10% in March 2018. The Containerd also achieves amazing growth, from 3% in March 2018 to 29% in early 2019.

CNCF also hosts four projects created in China, which are more widely used in China. Among them, Dragonfly, 17% of respondents used in production and KubeEdge, 11% of respondents used in production, are the two most commonly used Sandbox Level Projects. Now both are in incubation stage. Harbor and TiKV are graduated CNCF projects, with 27% and 5% of usage respectively in production.

The newly graduated project Rook is excluded.*

Newly incubated projects Argo, Contour, and Operator Framework are excluded. Rook is now a graduated project.*

Since CNCF’s last survey in China, the benefits of using cloud native projects in production have changed.

• Faster deployment benefits most for the first time, mentioned by 47% of respondents.

• Improved scalability remains in the second place as before, accounting for 35%.

• Cost savings still ranks third, accounting for 33%.

• Improving developer productivity, cloud portability and higher availability tied for fourth, with a proportion of 31%. Availability ranked first and portability ranked fourth in the survey of November 2018.

Serverless

In the survey in China, 36% of respondents use hosted platforms and 22% use installable software.

For enterprises that use hosted platforms, the top three suppliers are Alibaba Cloud Function Compute (FC) (46%), AWS Lambda (34%), and Tencent Cloud Serverless Cloud Function and Huawei FunctionStage (both occupy 12% of respondents).

For those users who use installable software, Kubeless ranks first, accounting for 29%, and is followed by Knative for 22% and Apache OpenWhisk for 20%.

In the 2019 survey in China, new questions about cloud native storage and service network are added. Both of them are popular cloud native projects that offer following advantages in active production environments.

Storage

The most commonly used cloud native storage projects are Ceph, accounting for 24%, Amazon Elastic Block Storage (EBS) for 23%, and Container Storage Interface (CSI) for 18%.

Service Network

Cloud Native Community in China

CNCF now has nearly 50 members in China. Evaluating by the number of contributors and submitters, China is also the third largest contributor to the CNCF projects, only behind the United States and Germany.

Here are some case studies of Chinese companies as below.

• JD.Com, Inc. has adopted Harbor as the central memory of its private images, saving about 60% of maintenance time.

• China Minsheng Bank has improved the delivery efficiency by 3 to 4 times, and its resource utilization has doubled through Kubernetes.

• Ant Financial Services Group has increased ten-fold in operations by using cloud native technologies.

Kubernetes and Cloud Native courses have been offered in China and more than 20,000 people have participated. Recently, the first China Cloud Native and Open Source Virtual Summit has been hold successfully as well.

The community in China is learning about cloud native technologies in many different ways.

Documentation

72% of Chinese respondents learn about cloud native technologies through documentation. There is a large number of documentation about each CNCF project on corresponding websites. For documentation of CNCF projects, see: https://www.cncf.io/projects/

CNCF invests thousands of dollars annually to improve all aspects of project documentation, including project documentation hosting, tutorials of adding documentation, and operation guides.

Activity

Participation in activities is a popular way for respondents to learn about cloud native technologies.

41% of respondents chose KubeCon + CloudNativeCon to learn new technologies. The next virtual KubeCon + CloudNativeCon is planned to be held from November 17 to November 20.

37% of respondents chose meetups and local events, such as Cloud Native Community Groups, as the way to learn about cloud native technologies.

Webinar

22% of respondents learn about cloud native technologies through technical webinars, while 8% choose business-oriented webinars, and other 8% choose CNCF virtual webinars.

CNCF has improved its webinar projects and planned to arrange regular webinars for Chinese audiences. Audiences can view the schedule, videos and slides of upcoming webinars, and playback of previous webinars through the website below https://www.cncf.io/webinars/