Best-in-Class Event Streaming
for Agents, Apps, and Analytics
for Agents, Apps, and Analytics
Sign up for our to stay updated.
Customer Story
Tencent Cloud replaced their Apache Flink + MySQL architecture with RisingWave for Quality of Service monitoring, achieving dramatically higher throughput while eliminating RocksDB storage costs.
The Challenge
Tencent Cloud's Quality of Service monitoring system relied on Apache Flink with MySQL for storage, but experienced significant TPS decline as data volumes scaled. Debugging multiway joins in Flink Java was complex and time-consuming, and mounting RocksDB storage costs added to the operational burden.
The architecture required maintaining three separate systems — Flink for stream processing, MySQL for serving queries, and RocksDB for state management. Every change required specialized Flink Java expertise, making the system difficult to evolve and expensive to operate.
The Solution
RisingWave unified the streaming, storage, and serving layers into a single system, replacing the three-component Flink+MySQL+RocksDB architecture. PostgreSQL SQL replaced Flink Java for development, and chained materialized views provided clear data lineage throughout the pipeline.
The system integrated natively with Tencent's existing infrastructure — Tencent Kubernetes Engine (TKE) for orchestration and Tencent Object Storage (TOS) for durable state. Chained materialized views replaced complex Flink DAGs with transparent, SQL-based data flow.
Single streaming database replaced the three-component Flink + MySQL + RocksDB architecture.
PostgreSQL-compatible SQL replaced Flink Java, eliminating the need for specialized streaming expertise.
Materialized views chain together for clear data lineage and transparent pipeline tracking.
Native integration with Tencent Kubernetes Engine for seamless container orchestration.
Tencent Object Storage provides durable state management without RocksDB overhead.
Eliminated RocksDB dependency entirely, removing associated storage costs and operational complexity.
Results
Tencent achieved a logarithmic leap in TPS performance, eliminated their RocksDB dependency entirely, and consolidated three systems into one. Debugging complexity was dramatically simplified through SQL-based development, and chained materialized views enabled clear data lineage tracking.
| Metric | Before | After (RisingWave) |
|---|---|---|
| TPS performance | Declining at scale | 10x+ improvement |
| Systems maintained | Flink + MySQL + RocksDB | RisingWave only |
| Development language | Flink Java | PostgreSQL SQL |
| State storage costs | Mounting RocksDB costs | Eliminated |
| Debugging complexity | Complex multiway joins | SQL-based, transparent |
| Data lineage | Opaque | Chained MV tracking |
Replace Flink + MySQL + RocksDB with a single streaming database.
Simplify Your Streaming Stack →