Nimbus: Baidu Search Platform's Low-Code Platform Practice and Exploration

According to Gartner research, application development demand growth exceeds IT delivery capacity by at least 5 times. It is estimated that by 2025, 70% of new application development will use low-code technology. This article explores how Baidu's Search Platform addresses the challenges of high-frequency requirement iterations and increasing innovation demands through low-code solutions.

The Search Platform provides two access methods: configuration-based customization using API interfaces, and code-based development within the platform through Search-AE (Application Engine). With 200+ independent business systems, the platform faces three main challenges: lack of effective knowledge沉淀, increased system complexity due to rapid iteration, and cumbersome full-process operations.

The solution employs a graph engine to drive business logic execution, where both general and customized capabilities are provided as operators. Businesses connect these operators through DAG graphs, enabling flexible system construction. The Nimbus low-code platform, built on iCoding IDE, provides five key capabilities:

1. One-click cloud development environment: Using Docker containers with Overlayfs for fast image pulling (5 seconds) and efficient dependency management through NFS mounting.

2. Visual operator orchestration: Operators are declared via annotations, scanned by the tool, and added to a warehouse for drag-and-drop business scenario configuration.

3. End-to-end debugging: Sandboxed environments replicate production modules, enabling efficient debugging without deployment overhead.

4. Modern testing tools: One-click deployment of simulation instances and automated regression testing with recorded traffic.

5. Intelligent capacity management: Automated analysis and decision-making for service deployment optimization based on QPS, latency, and resource load data.

The platform has achieved significant benefits across multiple business units, validating the feasibility of using low-code for cost reduction and efficiency improvement in complex business scenarios.