Design and Architecture of a High‑Performance Commercial Advertising Platform

The article introduces the advertising business model, describing the three main participants—publishers, advertisers, and users—and outlines the massive revenue potential of internet ads.

It then details the composition of a generic ad system, which consists of four core components: the serving platform, ad engine, algorithmic strategies, and data platform.

Microservice Migration : The transition from a monolithic architecture to a micro‑service oriented architecture (MSOA) follows three principles—divide, combine, and refine—resulting in a layered service hierarchy with API/Web entry points, a computation layer, and an infrastructure layer.

Service Governance : Effective governance requires RPC frameworks and capabilities such as communication, service discovery, monitoring, fault tolerance, contract management, continuous delivery, and security. The article lists concrete tools used (e.g., HSF, Pandora‑boot, Diamond, Sentinel, K8S, Aone).

Database Layer : Primary storage relies on MySQL InnoDB for OLTP workloads and Alibaba's X‑Engine for ultra‑high‑throughput transactions. The storage strategy includes sharding, master‑slave replication, multi‑region high availability, and consistency mechanisms based on Paxos.

Data Transmission Flow : After ads are stored, a real‑time binlog/DRC pipeline streams changes to the search system, enabling low‑latency candidate generation for user queries.

Real‑Time Billing : A Flink‑based billing engine processes media events, ensures exactly‑once semantics, and updates budget status in real time.

Reporting Platform : The reporting stack uses a combination of Kylin (MOLAP) and Alibaba Cloud ADB for minute‑level data, employing a pyramid model and pre‑aggregated cubes to support massive OLAP queries with sub‑second response times.

Optimization Techniques : The article highlights CBO optimizations such as TOP‑N deep pagination, bucket statistics, predicate push‑down, and parallel query execution to achieve scalable query performance.

Finally, the article summarizes the overall Alibaba technology ecosystem, emphasizing the integration of infrastructure, platforms, and business services to support a robust advertising platform.