Cloud Native Innovation Forum: AutoMQ Table Topic, OceanBase Integrated Database, and Observability Practices

In the digital era, cloud native technologies have become essential for rapid application deployment, high agility, scalability, and cost efficiency, driving enterprise digital transformation worldwide. Leading companies such as OpenAI and Nvidia leverage cloud native for compute scheduling, while Chinese policies also support its adoption.

Zhihu has been actively exploring cloud native since 2017, achieving full containerization, large‑scale distributed databases, service mesh integration, offline mixed deployment, APISIX gateway replacement, and deep involvement in AutoMQ development alongside OceanBase and Flashcat.

At the end of 2024, Zhihu co‑hosted a Cloud Native Innovation Forum with AutoMQ, OceanBase, and Flashcat, where experts presented multi‑angle discussions on practical cloud native applications.

AutoMQ Table Topic : The solution enables direct streaming data ingestion into storage, eliminating the need for separate Spark/Flink ETL pipelines. Data flows from Kafka API to a streaming storage layer, then synchronizes to table storage, supporting both online services and data warehouse analytics. Core components include schema registry, table coordinator, and table worker, which automatically evolve Iceberg tables and provide zero‑ETL, auto‑scaling, load balancing, and cost reduction.

The Table Topic is available in AWS Beta and will expand to other cloud platforms.

Zhihu's AutoMQ Cost‑Saving Practices : Facing Kafka scaling challenges, Zhihu evaluated messaging solutions and selected AutoMQ for its S3 stream architecture, reducing write latency, optimizing cold‑read performance, and storing data in low‑cost object storage. Extensive 24/7 stress testing, kernel tuning, and failure injection ensured stability at large scale.

Zhihu contributed over 35 patches to the AutoMQ community, built a Kubernetes operator for rapid cluster provisioning, and implemented automatic failover and load‑aware scheduling, achieving significant resource and operational cost reductions while increasing throughput.

OceanBase Integrated Database : Presented as a unified solution supporting both transactional (TP) and analytical (AP) workloads, with single‑node to distributed cluster scalability, multi‑cloud deployment, row/column storage, vectorized query engine, and three‑replica high‑availability architecture. Use cases include highly compressed historical databases, multi‑tenant MySQL consolidation, 24/7 high‑availability, HTAP, multi‑model KV, and AI‑driven vector search, with notable deployments at Kuaishou, Haidilao, and Industrial and Commercial Bank of China.

Observability + AI – Flashcat Practices : Flashcat's monitoring platform integrates microservice, network, and middleware metrics, providing end‑to‑end observability, anomaly detection via multi‑model fusion, and AI‑driven alerting and intelligent inspection. The system supports automated fault localization, risk prediction, and conversational bots for status queries.

Global Observability with Flashcat : For overseas services, Flashcat employs a central‑plus‑edge architecture, enabling closed‑loop alerts, multi‑source data integration, and 24‑hour reliability across multi‑region deployments.

The forum concluded with thanks to speakers and participants, promising more specialized technical sessions in the future.