Transforming Data Operations: How TiDB Revolutionized Trip.com’s Global Travel Business

Trip.com is a leading global one-stop travel platform, encompassing brands such as Trip.com, Qunar, and Skyscanner. Trip.com offers services to over 90 million members, including hotel reservations, reviews, special hotel searches, flight bookings, status and schedule inquiries, ticket price searches, and flight queries.

To handle its data volumes, which reach into the billions, Trip.com uses TiDB‘s HTAP capabilities. This boosts efficiency in areas like their international CDP platform, hotel settlement, and risk control.

The Challenge: Navigating Complex Data Integration for International Business Optimization

Trip.com faced significant challenges in its international business sector due to the complexity of managing diverse markets, products, and distribution channels, as well as the high costs associated with traffic acquisition. Key challenges included:

1. Data Collection and Management
   • Trip.com needed to manage and enrich a vast array of data sources, which reached into the billions. These included:
     • First-party data: Internal sources such as UBT logs, platform data, customer service processing data, and app installation data.
     • Second-party data: Data from other brands within the group, like SC and Travix.
     • Third-party data: Partner websites, such as meta-distribution platforms.
   • The diversity of data formats (structured, semi-structured, and unstructured) and the combination of offline and online processing added complexity.
2. ID Matching
   • Data from different sources came with different ID tags, such as unified ClientIDs for app data. Each business system had its own IDs and tags, with varying degrees of relationships between them.
   • The challenge was the need for a unique ID to link data across platforms. A comprehensive ID mapping was required to convert and link tags across different entities without a unified ID.
3. Business Tag Model
   • Trip.com needed to develop business tags for scenario-based decision-making, such as identifying the most popular products or travel destinations.
   • Initial efforts resulted in hundreds of statistical tags, but many were not directly useful, overwhelming product and operations teams.
   • The challenge was creating meaningful, prioritized tags aligned with specific business scenarios.
4. Tag Usage
   • Integrating these tags with various systems, including messaging platforms, third-party platforms, and in-house platforms, was crucial for maximizing their impact on business performance.
   • The challenge was ensuring the CDP could seamlessly integrate with these systems.
5. Data Processing Timeliness and Flexibility
   • The existing CRM data system operated on a T+1 data warehouse process, which led to slow data processing and poor timeliness. Data was often calculated offline and stored in the ES cluster.
   • Hundreds of tags were online, with over 50% in use, but they primarily served scenarios where data timeliness was not critical.
   • The challenge was to develop a new system that could dynamically configure data processing logic, consume message data, and update tags in real time to meet the demands of business systems, EMD, PUSH, and other use cases.

These challenges highlighted the need for Trip.com to develop a more robust, real-time data processing system that could handle the massive volumes of diverse data, improve operational efficiency, and support the company’s growth in the international market.

The Solution: Leveraging Real-Time Data Processing with TiDB

Below is the proposed solution structured according to the four main challenges:

1. Data Collection and Management
   • Enhanced Data Integration: The CDP platform is designed to enrich data from diverse sources—first-party, second-party, and third-party data—by supporting structured, semi-structured, and unstructured data formats. The system integrates internal logs, platform data, customer service data, app installation data, and partner website data.
   • Dynamic Data Processing: The platform utilizes a Kappa-like architecture for real-time data processing, ensuring immediate availability of processed data for business systems, such as EMD and PUSH. This allows for a more responsive and timely application of data in marketing and operational scenarios.
2. ID Matching
   • Unified ID System: A unique ID is proposed to link various data sources, allowing for seamless integration and utilization across different business systems. Where a unique ID does not exist, a comprehensive ID mapping system is implemented to enable the conversion and linking of different IDs.
   • Dynamic Rule Configuration: The system supports dynamic rule configuration for ID matching, allowing real-time modifications and immediate application without the need for restarting tasks. This ensures that business systems can combine and utilize IDs effectively in real-time.
3. Business Tag Model
   • Prioritized Tag Development: Instead of overwhelming product and operations teams with numerous tags, the system focuses on creating business tags that are directly relevant and effective for specific business scenarios. The system dynamically generates tags based on scenario-based decision-making needs, ensuring that only useful tags are provided.
   • Rule-Based Tag Processing: The platform utilizes a rule engine that supports basic and custom operators to process data dynamically. This includes operations like filtering, joining, and prioritizing data to create business-relevant tags in real-time.
4. Tag Usage
   • Real-Time Integration: The platform integrates with systems that utilize business tags, such as messaging systems, third-party platforms, and in-house platforms, through a dynamic real-time tagging processing system. This ensures that tags are immediately applied to enhance performance in marketing and operational activities.
   • OLAP and OLTP Query Support: The system supports both OLAP (for complex analytical queries) and OLTP (for real-time transactional queries) scenarios. This is achieved through the use of TiDB, which handles both real-time and batch data processing, ensuring the persistence and timeliness of business tags.

Trip.com’s platform utilizes a hybrid architecture to support its complex data processing needs. A modified Kappa-like architecture is implemented for real-time data ingestion and processing. This ensures that real-time triggers, such as App Push notifications and email marketing, are efficiently handled without intermediate data storage or querying. This approach allows the system to dynamically configure and process real-time business triggers, enhancing user engagement and conversion rates while minimizing development, testing, and deployment efforts.

For scenarios requiring the persistence of business tags, a Lambda-like architecture is employed. TiDB serves as the storage backbone, enabling seamless integration of batch and real-time data processing. This setup ensures that business tags are consistently up-to-date and accessible for various business applications, including CRM systems. The ability to handle both OLAP and OLTP queries with real-time, millisecond-level response times makes this architecture highly effective for supporting business-critical functions.

The Results: Transformative Results Achieved with TiDB Integration at Trip.com

The integration of TiDB into Trip.com’s data infrastructure has been a game-changer, providing the agility, speed, and scalability necessary to support the company’s complex and growing international operations. Here are the results in detail: 

1. Enhanced Data Processing Speed:
   • TiDB’s hybrid transactional and analytical processing (HTAP) capabilities have drastically reduced the time required for data processing tasks. This improvement has enabled Trip.com to transition from a T+1 data processing model to near real-time data handling, which is crucial for dynamic business environments.
   • The real-time data processing has allowed Trip.com to react promptly to changes in user behavior, improving the timeliness and relevance of targeted marketing efforts and other customer interactions.
2. Increased Query Performance:
   • TiDB’s support for both OLTP (Online Transaction Processing) and OLAP (Online Analytical Processing) queries has allowed Trip.com to handle complex queries with high concurrency efficiently. This dual capability ensures that both transactional workloads and analytical queries are processed without performance degradation.
   • The system’s ability to perform fast, millisecond-level query responses has enhanced user experience and system reliability, particularly in scenarios requiring immediate data retrieval.
3. Scalability and Flexibility:
   • TiDB’s distributed architecture has provided Trip.com with the scalability needed to handle massive data volumes across multiple business scenarios. As Trip.com’s international business grows, TiDB’s ability to scale horizontally ensures that the system can meet increasing demands without compromising performance.
   • The flexibility in data storage and query options—whether for real-time triggers or persistent tag data—has empowered Trip.com to tailor its data processing to specific business needs, allowing for more precise and effective decision-making.
4. Reduced Operational Costs:
   • By automating and streamlining data processing tasks, TiDB has reduced the operational overhead associated with maintaining separate systems for transactional and analytical workloads. This consolidation has not only lowered infrastructure costs but also reduced the complexity and time required for system maintenance.
   • The system’s improved efficiency has translated into better resource utilization, further cutting operational expenses and boosting overall performance.
5. Improved Business Outcomes:
   • The enhanced data processing and querying capabilities have led to better-informed business decisions, contributing to higher conversion rates, increased customer satisfaction, and, ultimately, more significant revenue. For instance, the ability to push real-time notifications based on user behavior has directly impacted customer engagement and purchase decisions.
   • The system’s reliability and speed have also enabled Trip.com to refine its marketing strategies, reduce churn, and improve customer retention, which are critical for maintaining a competitive edge in the global travel market.

Conclusion 

Trip.com’s use of TiDB has transformed its data operations, showcasing the power of a hybrid database solution. TiDB’s real-time processing, dual-query performance, scalability, and cost efficiency have enhanced Trip.com’s agility and growth. The future looks promising as Trip.com continues to leverage TiDB’s advanced features to stay ahead in the competitive global travel market.

As you explore opportunities to elevate your organization’s data infrastructure, consider how TiDB can be a game-changer for your business. 

Ready to see TiDB in action? Schedule a demo today to discover how our hybrid database solution can drive innovation, efficiency, and growth for your organization.