Yesterday SAP for $10M, Today Java Microservices for ₽10M
summary
A team replaced a SAP-based daily price-calculation service for 20,000 stores with Java microservices, making it 4× faster, ~70× cheaper and far more resilient while adding transparent scaling and load balancing. The talk describes the on-the-fly design and implementation of this business-critical, high-data-volume service, focusing on parallel-processing problems encountered, solutions used, profiling techniques that sped algorithms by hundreds-to-thousands×, memory-reduction tactics, and practical tips for handling tens-to-hundreds of millions of records.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 2:29Project goals
- 8:13Gradual load migration
- 10:18Data loading problems
- 14:57ClickHouse for loading
- 17:47Data reconciliation adapters
- 24:15Price calculation optimization
- 33:03Parallel price calculation
- 38:03Database read optimization
- 40:36Reconciliation orchestration
- 43:29Replication and results
- 46:20Conclusion
related talks
45mDesigning, building, testing and deploying microservices
43mCaching and Beyond - Smarter Data Processing with Java
31mSharing is caring: Building a modern microservice architecture helps everyone get ahead
47mSay Goodbye to Microservices, Say Hello to Self-Contained Systems
55mModernizing Legacy Code - While Not Bringing Down Prod
53mHigh Performance Java Cloud Native Apps
15mBehind the Price Tag: Architecting for a Billion Prices
37mMicroservices: Get Rid of Your DBA and Send the DB into Burnout