Key Insights from 2017 DevOpsDays Shanghai: DevOps Predictions, Continuous Delivery Cases, and Contract‑Driven Service Teams

Below is the content of my 2017 DevOpsDays Shanghai talk; more details can be found on the conference website.

Main topics include:

2011 DevOps prediction

Two‑stage case study of building continuous delivery capability

Relationship among Agile, DevOps and continuous delivery

Evolution of application deployment

Challenges of continuous delivery across three dimensions

The "continuous delivery puzzle"

Future: contract‑driven service teams

How microservice architecture expands developers' responsibilities

Contracts exist beyond service interfaces

1. 2011 DevOps Prediction

At the 2011 QCon Hangzhou, I said that "Agile" was a buzzword and that DevOps would become one in a few years – a prediction that has now come true.

2. Two‑Stage Continuous Delivery Case

While working as a PMO at Baidu, I led a pilot transforming a large‑search team from waterfall to continuous delivery. The first stage (Waterfall → water‑scrum‑fall) took 25 weeks to deliver, establishing the foundation of requirement splitting, trunk‑based development, CI, and automated testing.

The second stage shortened the release cycle from three weeks to two weeks by making DevOps a collaborative effort between developers and operations, introducing partner‑style involvement of ops and one‑click automated deployment scripts for development, test, and production environments.

3. Agile, DevOps and Continuous Delivery Relationship

A traditional software development flow passes a concept through four separate departments, creating “department walls”. Breaking these walls with agile practices and DevOps enables end‑to‑end delivery.

Continuous delivery is an organizational capability that spans people, processes, technology, and aims for high‑quality, low‑cost, low‑risk rapid delivery of business value.

4. Evolution of Application Deployment Operations

Deployment practices have progressed from manual scripts, to office‑automation forms, to version‑controlled scripts, to abstracted deployment models using tools like Puppet, Chef, Ansible, and SaltStack, and finally to ecosystem‑level operations driven by containers, cloud, and microservices.

Microservices simplify development (single‑responsibility services) but increase operational complexity, requiring new techniques to keep operations manageable while preserving flexibility.

Three dimensions affected by rising continuous‑delivery demands are infrastructure, software architecture, and organizational mechanisms.

5. Challenges in Three Dimensions

Infrastructure challenges (build, branch, test, release, monitoring) are the most visible and easiest to start addressing, but they quickly expose architectural and organizational difficulties that must be tackled in tandem.

6. Continuous Delivery Puzzle

The puzzle is illustrated with a traditional Chinese “seven‑piece” puzzle, symbolizing that each organization must assemble its own shape from the same set of pieces.

7. Future – Contract‑Driven Service Teams

The likely future is a contract‑driven, high‑cohesion, low‑coupling service‑oriented architecture, where both software and organizational structures are defined by contracts.

Service teams provide standardized infrastructure services (package management, image distribution, monitoring, etc.) to application teams, which in turn deliver business value while ensuring reliability in both production and test environments.

8. Expanded Responsibilities After Microservice Architecture

Beyond production stability, teams must also guarantee test‑environment reliability, because failures in a test service can block other teams' development.

9. Contracts Beyond Service Interfaces

Contracts should cover not only production stability but also the quality and stability of integration and test environments, ensuring clear boundaries and responsibilities across service‑oriented teams.