Community Blog DevOps Capability Improvement Model - Alibaba DevOps Practice Part 26

DevOps Capability Improvement Model - Alibaba DevOps Practice Part 26

Final part of this 27-part series discusses the importance of DevOps capabilities as an organization's size and complexity grow.

This article is from Alibaba DevOps Practice Guide written by Alibaba Cloud Yunxiao Team

DevOps capabilities reflect the ability of technology R&D to respond to business changes. DevOps capabilities become increasingly important as an organization's size and business complexity grow. Continuous improvement in DevOps capabilities is a common challenge to technology R&D.

We have defined the DevOps capability maturity model to indicate the direction for organizations' DevOps capability improvement and plan a clear path. The model offers two values:

1) To know where we are today

2) To know how to plan the improvement path

We have divided DevOps implementation into four categories and ten capabilities:

Basic capability includes the service level and infrastructure level of the system, which are the foundation of R&D and delivery. Among them, the service level is closely related to the application architecture. Serverless architecture is ideal. Rock architecture is a lower-level one that couples with the entire system. The infrastructure level reflects the attention of R&D required for infrastructure. Higher attention results in higher costs of R&D on infrastructure while the lower the efficiency. The stability cannot be guaranteed.

Delivery capability includes tool, test automation, and deployment and release levels. It is a typical indicator of engineering development. The tool level refers to the overall level of tools involved in the whole R&D process, including single-point capabilities (such as project collaboration tools, building tools, dependency management tools, and environment management tools) and coordination capabilities (such as the connection between requirements and release systems, defects, and tests). Test automation level refers to the feedback efficiency and automation degree of the test. Test automation is an important part of engineering capabilities and the foundation for improving deployment and release capabilities. Deployment and release level is the ability to launch products into the production environment and provide services, including the automation level, stability (such as smooth phased release), and adaptability (the ability to handle different scenarios and recover) of releases. Good delivery should be continuous and fast with high quality and low risk.

O&M capability includes system observability, application O&M level, and infrastructure O&M level. It reflects the level of elasticity and scalability during system runtime. Observability is the most important part and is mainly reflected while seeing the global operations and problems from the perspective of the system. It is usually reflected in monitoring, process analysis, and problem locating. Application O&M refers to the O&M operations performed on applications, such as configuration item modification, application runtime parameter adjustment, and scaling. Infrastructure O&M refers to the O&M operations on the infrastructure of the system, including virtual machines, container platforms, and basic services (such as domain names and a configuration center). It is the basic part of the entire system. The best O&M is automatic O&M.

Collaboration capability includes the collaboration between business and technology and development and O&M. It reflects the overall collaboration and business response capabilities. The collaboration between development and O&M aims to make the delivery process smoother and more efficient, improve technical response speed, and ensure the flexibility and resilience of system operations. The collaboration between technology and business aims to make the value transfer and delivery from business to technology more accurate and efficient and made feedback more immediate. Thus, it can improve the efficiency and effectiveness of business development and innovation.

Maturity Model

Based on ten capabilities in four categories, we provide a maturity model with five levels. The maturity of each level increases progressively from L0 to L4.


L0 features manual batch delivery and manual O&M. This is a DevOps stage with zero capability. Its service capability depends on each developer. The quality of service delivery is generally not high. Problems emerge as the business grows and the size of the team changes. Assistance from relevant tools is needed at first.

L1 features manual-based and tool-assisted batch delivery and O&M. This stage introduces automated tools to assist with O&M and release. Generally, the foundation for servitization is prepared, and part of the infrastructure has already been migrated to the cloud. Some open-source tools are introduced or created to meet specific requirements. However, these tools are often isolated. Business, development, and O&M synchronize regularly, and requirements are delivered in batches.

L2 features partially automated delivery and O&M based on business requirements. This stage makes continuous delivery possible based on business requirements. Declarative O&M is adopted, and cloud-native infrastructure is usually used. In addition, cloud resources are used to manage the service status, and most toolchains have been combined to achieve a certain degree of continuous delivery. Services begin to have middleware-level abstraction and governance capabilities. However, automated O&M is not available in general yet. Operations, such as rollback, still need to be judged and processed manually.

L3 features end-to-end automated delivery and constrained automated O&M based on business requirements. This stage improves the delivery frequency and quality of business requirements significantly. Also, the service level has been quite high. In most cases, business development can focus on specific technology stacks, and major services are released and run in serverless mode. The release process can be automated and declarative, with only a few interventions required for gray release. Services can be automatically maintained and governed in most cases.

L4 features end-to-end continuous delivery and closed-loop adjustment of business requirements with completely automated O&M. Developers only need to focus on business development in this stage, and business requirements can be delivered and adjusted quickly. The service level is decoupled from the technology stack, being fully serverless. The entire delivery process is fully automated, and services can be governed automatically. This stage is our goal.

0 0 0
Share on

Alibaba Cloud Community

900 posts | 201 followers

You may also like