How Victoria's Secret Migrated 2 Billion+ Documents with Custom Bidirectional Replication скачать в хорошем качестве

How Victoria's Secret Migrated 2 Billion+ Documents with Custom Bidirectional Replication

Learn why more than 50,700 customers have chosen to build their applications on MongoDB here: https://trymongodb.com/3MzXwWN Subscribe to MongoDB YouTube→ https://trymongodb.com/3yY7RZH Discover how Victoria's Secret successfully migrated over 2 billion documents by engineering a custom bidirectional replicator between two different databases (CouchDB and MongoDB). This session covered the strategies used to maintain zero downtime while supporting a platform with close to 3 million DB Transactions. Learn about the performance and scalability improvements achieved, the challenges faced, and the solutions implemented and key learnings from this complex migration process. Speaker: Hemanth Kumar Vemola, Lead Platform Engineer, Victoria's Secret & Co ⏱️ Timestamps ⏱️ Introduction and Migration Overview 00:00:00 The speaker begins by greeting the audience and expressing excitement for the MongoDB event in Austin. They introduce HHemanth Kumar Vemola, the lead platform engineer from Victoria's Secret, who will discuss the successful migration of over two billion MongoDB documents from CouchDB to MongoDB Atlas with a custom bidirectional replication process, resulting in zero downtime and improved scalability and performance. Victoria's Secret E-commerce Platform Evolution 00:05:06 Hemanth Kumar Vemola provides background on Victoria's Secret and its transition from a monolithic e-commerce platform to a microservice architecture. He explains the initial choice of CouchDB as their NoSQL database and the challenges faced, including data duplication, lack of support, and operational complexity during cloud migration efforts. Migration Challenges and Strategy 00:10:09 The speaker outlines the expectations and requirements for the migration, such as zero downtime and real-time data replication. He discusses the challenges of replicating data between MongoDB and CouchDB, making schema modifications, and ensuring no disruptions to ongoing projects. Migration Execution and Phases 00:15:03 The migration process is broken down into phases, with the initial phase focusing on one database from each of five categories to predict challenges. Subsequent phases applied these learnings, with the final phases running concurrently to complete the migration of all databases. Post-Migration Optimization and Advanced Features 00:20:08 After the migration, the team focused on operational improvements, scalability, and cost optimizations. They also began to leverage advanced MongoDB Atlas features such as search, charts, and online archiving. Q&A and Final Thoughts 00:25:04 The speaker answers questions from the audience, sharing insights on the migration journey, including the use of Go as a programming language, schema modifications, pre-scaling for high traffic events, and conflict resolution between database writes. The session concludes with a summary of the migration's impact and learnings.