From Scripts to AIOps: How Alibaba’s Ops Evolved and What Skills You Need Today

Reviewing Alibaba’s operations development stages—from early manual/script Ops, through simple tools and automation, to systematic platform‑based DevOps, and finally exploring intelligent Ops—highlights how data fuels AIOps, the importance of DataOps, and the evolving skill requirements for Ops professionals.

Human/Scripts Ops

Operations work demands broad, integrated skills. Early perceptions limited Ops to deployment, change, alert handling, and machine movement. In small companies, engineers solved diverse problems quickly by copying solutions from Google, Baidu, forums, and blogs, often without deep understanding of underlying principles or system‑wide context.

Tools Ops

Ops engineers love writing scripts for batch deployment, cleanup, and interactive guides, leading to duplicated tools, inefficiency, and security risks. Numerous incidents stemmed from simple mistakes such as mistyped characters. The industry recognized the need to consolidate scripts into reusable tools, marking the early DevOps stage before full development integration.

Platform‑centric DevOps

As companies scale, the number of applications grows from a few to thousands or hundreds of thousands, and hardware expands from simple CPUs to GPUs, FPGAs, ASICs, and other heterogeneous platforms. Software architectures become complex with big‑data and distributed systems. Manual operation becomes impossible; Ops responsibilities shift to global architecture planning, resource operation and cost optimization, automation platform development, stability assurance, massive data analysis, and more.

DataOps (Data‑driven Ops)

Increasing business demands require not only breadth but depth in specific areas. By applying software engineering and data‑driven approaches, a comprehensive intelligent Ops toolchain is built to manage ultra‑large distributed clusters, improving stability, efficiency, and cost, thereby raising the skill challenge for Ops personnel.

AIOps and the Road Ahead

While many hype AIOps as a solution that only needs powerful algorithms, the reality is that engineering maturity determines whether the algorithmic kernel can be effectively applied. Alibaba views the current stage as DataOps, using data to drive Ops and engineering implementation. The analogy to autonomous‑driving levels illustrates the difficulty of achieving full automation.

With big‑data growth, Ops engineers must acquire composite abilities: architectural design, development, deep Ops knowledge and business understanding, basic engineering algorithms, and technical project management (TPM).

Ultimately, AIOps must be embedded in operational platforms and products, requiring continuous human involvement for scenario definition, visualization, machine‑learning, and big‑data analysis to deliver lasting business value.

The transition brings organizational change: moving from a maintenance‑focused department to an R&D‑driven innovation unit, shifting mindsets from reactive firefighting to proactive, value‑driven, intelligent operations.