Best-in-Class Event Streaming
for Agents, Apps, and Analytics
for Agents, Apps, and Analytics
Sign up for our to stay updated.
Customer Story
CVTE is a $30 billion electronics manufacturer. They replaced their crash-prone in-memory system with RisingWave, enabling 10+ way streaming joins with sub-second freshness.
The Challenge
CVTE initially used PipelineDB and later built a custom in-memory stream processor. Both systems were single-instance only, required full recomputation on crash, consumed over 1TB of memory for complex multi-way joins, and offered poor observability into pipeline health.
No horizontal scaling was available to handle growing manufacturing data volumes, and every crash or restart meant rebuilding the entire pipeline state from scratch -- an unacceptable risk for production manufacturing systems.
The Solution
RisingWave introduced persistent checkpoints that eliminate recomputation on failure, decoupled compute and storage for independent scaling, Debezium CDC for seamless database ingestion, and native Grafana integration for comprehensive monitoring of all streaming pipelines.
State is stored durably rather than in memory, dramatically reducing resource requirements. The decoupled compute-storage architecture enables CVTE to scale compute and storage independently as their manufacturing data volumes grow.
Results
CVTE now runs 10+ way streaming joins with sub-second data freshness. The system eliminated over 1TB of memory requirements by using durable state storage, recovers from crashes without recomputation, and provides full pipeline observability through Grafana dashboards.
| Metric | Before | After (RisingWave) |
|---|---|---|
| Concurrent stream joins | Limited | 10+ way joins |
| Data freshness | Minutes | Sub-second |
| Memory required | >1TB in-memory | Durable state storage |
| Crash recovery | Full recomputation | Persistent checkpoints |
| Architecture | Single-instance | Horizontally scalable |
| Pipeline observability | Poor | Full (Grafana) |
Handle complex streaming joins without the complexity.
Scale Your Manufacturing Data →