This post was originally posted over on my personal website. Maybe you want to check it out [HERE].
MongoDB vs MySQL head-to-head: We benchmarked both databases with real Disneyland data. MongoDB won writes by 67% and some queries by 98%, but MySQL had surprises too!
When choosing a database for your next project, performance benchmarks often feel abstract and disconnected from real-world scenarios. That's why our recent university project comparing MongoDB and MySQL using actual Disneyland visitor data provides such valuable insights – we tested both databases against the same complex dataset with realistic use cases.
The Dataset: Perfect for Database Testing
We used a comprehensive Disneyland Paris dataset that presented real analytical challenges:
- Wait times for 37 attractions (2018-2019)
- Daily attendance data (2018-2022)
- Weather conditions (1999-2022)
- Park operational schedules (2018-2022)
This dataset was ideal for performance testing because it required:
- High-frequency writes (weather data every hour, wait times every 15 minutes)
- Complex aggregations across multiple data sources
- Time-series analysis
- JOIN operations between related data
Database Setup: Fair Fight Conditions
To ensure a fair comparison, we:
- Identical hardware: Same server (20GB RAM, 6 CPU cores, HDD storage)
- Equivalent indexing: Created the same indexes on both systems
- Standardized schema: Normalized data structure for MySQL, embedded documents for MongoDB
- Same queries: Translated identical business logic between SQL and MongoDB aggregation pipelines
Schema Differences
MySQL: Traditional normalized approach with separate tables for weather descriptions, linked via foreign keys.
MongoDB: Document-based approach with embedded weather objects, eliminating the need for joins.
Performance Results: The Numbers Don't Lie
We ran each query 100 times sequentially to get reliable averages. Here are the results:
Write Operations (Data Insertion)
| Operation | MySQL | MongoDB | MongoDB Advantage |
|---|---|---|---|
| Weather Data Insert | 56.7ms | 19.6ms | 65.4% faster |
| Wait Times Insert | 59.2ms | 19.3ms | 67.4% faster |
| Attendance Insert | 57.2ms | 19.3ms | 66.2% faster |
| Schedule Insert | 57.7ms | 28.4ms | 50.7% faster |
Takeaway: MongoDB consistently outperformed MySQL for write operations, often by more than 2:1 ratios.
Read Operations (Analytics Queries)
| Query Type | MySQL | MongoDB | Winner |
|---|---|---|---|
| RQ1: Average wait times by attraction | 52.6 seconds | 1.0 seconds | MongoDB (98.1% faster) |
| RQ2: Attendance by weather conditions | 413ms | 20.7 seconds | MySQL (98% faster) |
| RQ3: Wait times vs weather correlation | 1h 12m 37s | 5ms* | MongoDB |
| RQ4: Holiday vs regular attendance | 66.3ms | 58.5ms | MongoDB (11.7% faster) |
*MongoDB query completed quickly, MySQL query failed after 4+ hours
What These Results Actually Mean
MongoDB's Strengths Revealed
Simple Aggregations: The 98% performance advantage in RQ1 shows MongoDB's aggregation pipeline excels at straightforward grouping and averaging operations.
Complex Time-Based Queries: RQ3's dramatic difference (MySQL took over an hour, MongoDB completed in milliseconds) demonstrates MongoDB's superiority for complex temporal correlations.
Write-Heavy Workloads: Consistent 50-67% faster write performance makes MongoDB ideal for real-time data ingestion scenarios.
MySQL's Surprising Victory
RQ2 (weather-based attendance analysis) was 98% faster in MySQL. This query required complex temporary table operations and multiple data source correlations – an area where SQL's mature optimization still shines.
The Technical Why Behind the Performance
MongoDB Advantages:
- No JOIN overhead: Embedded documents eliminate expensive table joins
- Optimized aggregation pipeline: Purpose-built for analytical workloads
- Flexible indexing: Better suited for document-based queries
- Schema flexibility: No rigid structure constraints during writes
MySQL Advantages:
- Mature query optimization: Decades of SQL optimizer improvements
- Efficient temporary operations: Better handling of complex intermediate results
- Memory management: Superior for certain multi-table operations
Real-World Application Insights
Based on our findings, choose:
MongoDB when you have:
- High-frequency data ingestion (IoT sensors, real-time analytics)
- Time-series data analysis
- Simple to moderate aggregation needs
- Schema evolution requirements
MySQL when you have:
- Complex multi-source data correlations
- Heavy use of temporary tables and complex JOINs
- Mature application ecosystems expecting SQL
- Strict consistency requirements
The Unexpected Learning: Context Matters
The most valuable insight wasn't that one database is universally better – it's that workload characteristics determine the winner. MongoDB dominated write operations and simple aggregations by massive margins, while MySQL excelled at complex relational operations.
Performance Optimization Lessons
Index Strategy: Both databases benefited equally from proper indexing on date fields and frequently queried attributes.
Query Design: MongoDB's aggregation pipelines required different thinking than SQL, but often resulted in more maintainable code.
Hardware Considerations: Even on modest hardware (HDD storage), the performance differences were dramatic, suggesting they'd be even more pronounced on modern SSD systems.
Bottom Line for Database Selection
Don't choose a database based on hype or familiarity alone. Our Disneyland data analysis proved that:
- Write-heavy applications: MongoDB provides substantial advantages
- Complex analytical queries: Results vary dramatically by query type
- Mixed workloads: Consider using both databases for different use cases
The 98% performance differences we observed aren't edge cases – they represent real-world scenarios that could mean the difference between a responsive application and one that frustrates users.
This performance analysis was conducted as part of a Modern Database Systems course at TH Köln. The complete benchmark code and query implementations are available here for those interested in replicating these tests.
Top comments (0)