Social Media
Background
A major online media company manages vast data across scenarios like real-time public opinion monitoring, content verification, and multidimensional analysis. As data grows, they face significant challenges, including handling high concurrency during peak traffic periods, ensuring the accuracy of data and insights, and maintaining high performance. To overcome these hurdles, the company urgently requires a unified solution that can efficiently manage these complexities while providing the scalability and flexibility needed for future growth.
Customer Pain Points
High recall rate requirements
The business relies on high recall rates to ensure accurate data analysis. A recall rate below 98% leads to inaccurate insights, affecting overall business precision.
Lack of real-time data access
Immediate data retrieval upon ingestion is essential for real-time monitoring in public sentiment analysis. Without this capability, the business risks missing critical insights.
Limited scalability for high concurrent reads/writes
The current system struggles to support the high concurrent demands of tens of thousands of QPS and 2000 TPS while maintaining a 99% recall rate, impacting performance during peak loads.
Inefficient data management
The business primarily relies on the last 7 days of hot data for vector searches, but without proper data lifecycle management, handling and accessing this data becomes inefficient.
