BKJIA comprehensive report] Google's advertising system used MySQL as the backend database. Currently, some systems have been migrated to F1, and the proportion data is unknown. However, a core advertisement business has been migrated to F1. F1 features are not as fast as racing cars. To support the following features, F1 has a response time requirement. However, the response time can still be accepted by OLTP services.
1. Structured Data Storage; 2. Parallel SQL Execution engine; 3. General transaction support; 4. triggers; 5. Indexes supported; 6. Dynamic Scaling; 7. Cross-Data Center synchronization.
Appendix F1 Data Processing Flowchart
Compared with MySQL's low-latency writing, F1's strong consistency and storage system are more attractive. After the successful migration to F1, Google's advertising business core will enrich the customerfacing application suite, so that the entire system has no downtime, how to restructure the architecture and application, and hide the latency from external users to the greatest extent, will be mentioned in future articles. F1's dispersibility also allows it to easily expand and supports higher batch processing workload throughput than traditional RDBMS.
Thanks to red7, a netizen from TechMiMi.