Tiki.vn - How we scale as a tech startup
Download as PPTX, PDF73 likes22,646 views
This document discusses how a Vietnamese tech startup called Tiki scales its e-commerce platform and operations. It outlines the company's products, challenges with scaling, and technology stack including PHP, MySQL, MongoDB, Redis, Elasticsearch, and more. It also describes how the company prepares for "super events" that cause spikes in daily traffic, like online shopping holidays, through load testing, caching, and ensuring its platform and products can scale to meet high demand.
Tiki.vn - How we scale as a tech startup
- 1. HOW WE SCALE AS A TECH STARTUP Son Tran – Founder & CEO Tung Nguyen – Head of Technical
- 2. A little about us Giao hàng 24h
- 3. Our Products TALA - Tiki E-commerce Platform TOS - Tiki Operation System CMS, Order fulfillment, Delivery, Inventory Management,… Tiki Mobile App Tiki.vn Responsive Web E-Book store
- 4. Mobile Apps Tiki App Miki App
- 7. Challenges
- 8. • Minimum Viable Product • Customer-oriented • Refactoring • Don’t build everything • Be friendly with SaaS, PaaS (Platform), IaaS (Infrastructure) • Data-driven Our Philosophies
- 11. Fastest PHP framework Technology Stack - PHP Phalcon Framework Reality - Less than 100ms response time More focus on business layer
- 12. • Problem: • Lots of Product attributes • JOIN too much • Solution: • Indexing with NoSQL (MongoDB) • Transactions with MySQL • Scaling with MySQL Replication Technology Stack - Database MySQL EAV - Entity-attribute value Model MongoDB - Hierarchical Model
- 13. • Problem: • Real-time data changes updating • Availability, recovering from disaster • Solution • Cache map • Cache invalidator and builder • Use persistent • Engine: Redis Technology Stack - Caching How cache map works
- 14. • Problems: • Real-time indexing • Data aggregation (Facets) • Performance and Scaling • Solution • ElasticSearch + MongoDB River Technology Stack – Search Engine
- 16. What’s a super-event? Daily traffic Super-event peak 1. Hard to predict 2. Different behaviors 3. Investment cost
- 17. • August - Dzựt Cô Hồn Online • November - Hội Sách Online • December - Black-Friday, Cyber Monday, Online Shopping Day Provide incentives for customers to go online shopping more Tiki Super Events
- 18. Event Dzựt Cô Hồn
- 19. • 2014 • 28K Likes/Shares • 40K Registered users • 300K Dzut actions • 5K Requests/s • 2015 • 67K Likes/Shares • 150K Registered users • 1.5M Dzut actions • 20K Requests/s • Case-studies We had some attentions
- 20. Also from our companions
- 21. We survived
- 22. HOW?
- 23. A large and scalable platform
- 24. Hundreds times of load testing Powered by
- 25. And also from Product Design
- 27. About Us Or send your feedback to tung@tiki.vn and get a FREE coffee
Editor's Notes
- #5: Chia sẻ số liệu
- #9: Dẫn dắt: Product/Tech team luôn luôn phải đối mặt với cả núi TO-DO cần giải quyết, trong khi đối thủ cạnh tranh thì nguồn lực dồi dào Bài toán luôn phải giải quyết là Làm những việc thực sự cần nhất với nguồn lực hữu hạn. Priority: - Có lợi cho khách hàng thì làm - MVP – Thử làm liền - Data Driven – Nguyên tắc lựa chọn giải pháp - Refactoring: Không nhầm lẫn với việc làm nhanh nhưng ẩu => Refactoring liên tục, được remind bằng hệ thống (issues tracker) và con người (QC/PM/Team member)
- #10: Sửa lại slide, mô tả mục đích từng icon