How CVTE Achieved Real-Time Transformation in Manufacturing With RisingWave
RisingWave has become an integral part of CVTE’s infrastructure, managing numerous materialized views and enhancing real-time dashboard capabilities. Its architecture effectively handles complex streaming database queries, including more than 10-way streaming join, temporal filtering, aggregation, and so on.
CVTE is a global leading enterprise in the manufacturing industry, renowned for its expertise in consumer and business LCD display technology. Founded in December 2005, CVTE currently holds a market capitalization of $30 billion and has several subsidiary business companies.
The company’s primary business involves the design, development, and sale of LCD display control boards and interactive smart panels, which are widely used in sectors such as home appliances, educational information technology, and enterprise services. CVTE is dedicated to enhancing user experiences through product innovation, research and development, and continuously creating value for customers and users. Since its establishment, the company has leveraged its expertise in audio-video technology, signal processing, power management, human-computer interaction, application development, and system integration in the field of electronic products to innovate and develop products for various application scenarios.
Situation Before Using RisingWave
As CVTE’s business expanded, the number of internal systems increased, resulting in more complex data interactions between systems and higher demands for real-time data. Starting in 2019, CVTE introduced PipelineDB to address real-time production capacity computation-related applications. Subsequently, they also adopted KsqlDB and began evalute a commercial stream processing system X in 2021 to address real-time data collection scenarios in MRP (Material Requirements Planning) computation. Later on, the evaluation gradually applied to production environment for supply chain, sales, and financial systems, enabling these business systems to access real-time information related to materials, bills of materials (BOM), sales orders, production orders, inventory, and shipments. Based on the evaluation result, they found that system X can help resolve issues related to the timeliness of data computation, database performance, and data redundancy.
While system X met CVTE’s business needs, it cannot be widely adopted in production because it presented some core challenges and they are eager to seek a better stream processing library as a replacement.
- Stability: One of CVTE’s primary concerns was that back then system X only supported single-instance deployment and had an all-in-memory architecture. In case of system bugs or crashes, the views couldn’t be incrementally computed, requiring a full computation, resulting in long processing times and significant business impacts.
- Resource Costs: CVTE sought a new stream processing library that supported persistence to reduce memory consumption. System X’s all-in-memory architecture was demanding on memory resources, especially since many of CVTE’s real-time computed views had millions of records and involved over 10 stream joins, requiring high memory resources, some with memory configurations exceeding 1TB.
- Observability: CVTE found it challenging to effectively monitor its stream jobs. Issues arose when assessing source throughput, tracking System X’s job creation progress, and monitoring CPU and memory usage via system monitoring (M) for metrics. For example, it was difficult to track progress during view creation and the historical time it took to create views.
Why Choose RisingWave?
Dissatisfied with their existing system, CVTE began searching for a robust alternative that is capability of handling complex streaming queries and suitable for production deployment. RisingWave caught their attention. This advanced distributed streaming database offered a feature set that addressed the gaps in their previous system:
- Reliability: RisingWave featured persistent and consistent checkpoint. This not only increased the availability of stream jobs (allowing them to recover immediately after a cluster restore) but also simplified maintenance. With the reliable checkpoint mechanism of RisingWave, engineers no longer needed to worry about full re-computations during job recovery.
- Scalability: The platform adopted a decoupled compute-storage architecture, making it seamless and efficient to expand. For example, CVTE could conveniently expand their computing resources without impacting storage resources.
- Efficient Joins: RisingWave excelled in providing stable support for multiple stream joins. Considering CVTE needed to perform streaming join on tables from different databases, often involving more than 10-way join, they required a system capable of handling this intensity. RisingWave supported more than 10-way streaming joins in a stable way with freshness at the sub-second level.
- Observability: RisingWave brought improved observability to stream jobs. Firstly, the platform provided a range of metrics at different granularities via Grafana dashboards, making it suitable for continuous monitoring. Secondly, the internal state of stream jobs was queryable through SQL, which was immensely helpful for debugging. Moreover, as stream jobs were materialized views in RisingWave and its SQL was compatible with PostgreSQL, tools like DBeaver could be used directly, significantly simplifying SQL debugging and troubleshooting data-related issues.
- Technical Support: CVTE was highly satisfied with RisingWave’s ongoing and fast customer support. The RisingWave team remained active to CVTE’s requirements and provides prompt assistance with deployments and troubleshooting. Additionally, they excelled in quickly delivering high-priority feature enhancements and fixes.
Implementation of RisingWave at CVTE
CVTE’s interaction with the RisingWave team began about a year ago. Over the last one year, RisingWave continuously provides timely support and deliver key features based on customer feedbacks.
Today, RisingWave has become an integral part of CVTE’s infrastructure, managing numerous materialized views and enhancing real-time dashboard capabilities. Its architecture effectively handles complex streaming database queries, including more than 10-way streaming join, temporal filtering, aggregation, and so on. The system efficiently processes data through Debezium Change Data Capture (CDC) and has excellent capabilities for sinking data to downstream databases via JDBC and Kafka. CVTE’s technical team interacts with RisingWave daily using the DBeaver UI.
Looking ahead, CVTE plans to expand the use of RisingWave, exploring new use cases to maximize the potential of this partnership.
CVTE has a strong need to bring real-time data analysis in prodcution and has previously the evaluated stream processing system like PipelineDB, KsqlDB and commercial stream processing system X. However, due to issues with stability, observability, and performance, CVTE turned to RisingWave, an advanced distributed streaming database.
RisingWave offers strong reliability, scalability, efficient joins, outstanding observability, and exceptional customer support. The journey of CVTE adopting RisingWave highlights the importance of aligning technological solutions with operational requirements. Through its innovative features and unwavering customer support, RisingWave not only addresses the pain points encountered previously but also brings CVTE’s real-time data analysis operations to the next level. This transformation underscores the critical importance of adaptability and the pursuit of the best solutions when faced with evolving business demands.