How HuoLala’s YunTai BI Platform Transforms Data Visualization at Scale

1. What is HuoLala YunTai

HuoLala YunTai is the internally developed Business Intelligence (BI) visualization platform.

By December 2022, HuoLala operated in 352 Chinese cities with 660,000 active drivers and 9.5 million active users. Rapid business expansion created a strong demand for data analysis and value conversion, prompting the development of a self‑built BI platform.

2. System Architecture Overview

Business Intelligence (BI) originated in 1865 and now refers to using technology to turn data into knowledge that supports decision‑making and discovers commercial value. Core BI functions include data warehousing, ETL, analysis, mining, and visualization.

BI differs from simple data processing by providing cross‑business, cross‑system insights that drive intelligent business construction, relying heavily on frameworks, models, and compute power.

The overall data architecture is divided into two main layers:

Foundation & Platform Layer : Provides data storage, processing, and development capabilities through data warehouses and compute platforms.

Service & Application Layer : Focuses on data value mining, leveraging the lower layers to maximize data asset value.

Additional security measures include data permission control, security auditing, and data masking/encryption.

3. Design Goals and Usage Flow

The goal is to enable users to create professional, scenario‑based reports within minutes.

Typical workflow: add data source → create data model (configuration or custom SQL) → build visual reports and dashboards → publish with shared access and permission control.

Report creation uses the open‑source Apache ECharts component; users drag chart components and data fields to assemble reports.

4. Data Source Integration

YunTai connects to internal data sources such as MySQL, PostgreSQL, Phoenix, Doris, Hive, as well as Excel/CSV files and internal quick‑report/metric libraries. Users do not need to know connection credentials, simplifying permission management.

5. Data Modeling

Two modeling approaches are supported:

Model Configuration : A zero‑code, drag‑and‑drop method suitable for non‑technical users, abstracting differences between data sources.

Custom SQL : Allows flexible, advanced modeling with full SQL control.

Fields are automatically classified as dimensions or measures, with support for aggregation functions (SUM, COUNT, AVG, MAX, MIN, etc.) and column‑level or row‑level permission controls.

6. Report and Dashboard Creation

Fourteen chart types are available, including metric cards, funnel charts, cross‑tables, scatter plots, and stacked charts. Users bind dimensions and measures via drag‑and‑drop, and can configure sorting, totals, grouping, top‑N, and period‑over‑period analysis.

Dashboards are built by dragging prepared charts and filter components onto an interactive canvas, enabling self‑service analysis.

7. Visualization Technical Design

Data Source Connection : Unlike many commercial BI tools that require users to input credentials, YunTai abstracts connection details, allowing users to specify only the database and table, which simplifies permission management.

Flexible Queries : Supports complex SQL WHERE clauses, data type conversion, date aggregation, dimension grouping, sorting, drill‑down, and filter inter‑linking.

Data Calculation : Operations include field expression conversion, date aggregation, dimension grouping, totals, Top‑N, and in‑memory secondary calculations for results like cross‑tables.

Performance Optimization includes:

Source data caching (e.g., using H2 for preview data).

SQL result caching with Redis.

Cache for filter dropdown values.

Connection pool management to handle high concurrency.

Optimized cross‑pivot queries by re‑using previous page results.

Interrupting long‑running big‑data queries via Google Concurrent SimpleTimeLimiter.

Asynchronous multi‑threaded operations such as data export.

8. Conclusion and Future Plans

The platform is still in an internal incubation phase, with areas for improvement such as query speed, workflow simplicity, and user prompts. Future work focuses on two main aspects:

System Stability Governance : Enhancing monitoring, alerting, stress testing, and code robustness.

Code Design Integration : Balancing performance and readability, guided by references like “Clean Code” and “Design Patterns”.

Overall, YunTai aims to become a cost‑saving, efficiency‑boosting tool for the enterprise.