48

Gengo Jaws Days Tokyo 2014 Presentation

Embed Size (px)

DESCRIPTION

Scaling at Gengo Presentation at AWS Jaws Days Tokyo. Presented and outlines Derek Szydlowski. Assistance from David Gilbert, Yosuke Tomita, Andrea Belvedere and Matt Romaine

Citation preview

  • 1.Who we are Andrea Belvedere VP of Engineering Derek Szydlowski Lead Operations Engineer Yosuke Tomita Backend and Ops Engineer

2. Communicate freely. Connect to people-powered translation at scale. Gengo engineering 12 engineers, 9+ languages spoken 3. What is Gengo Submit your translation via Website or API 8000 translators and counting Language pairs and counting 54Web API 4. The Order Flow 5. Order translation Translator gets job Receive your translation 1 2 3 6. Order translation Translator gets job Receive your translation 1 2 3 7. Order translation Translator gets job Receive your translation 1 2 3 8. The API Fully automated workflow, no project management, plug into 8,000 translators A direct connection Client libraries, developer support and a friendly testing environment Quick to start Documented REST API makes integrating straight forward Clean and easy 9. Simple? 10. Complex Job Flow T T T T C C C C C C C C C C C C C C C Receive Translate Deliver 11. Scaling Concerns Translators Engineering Translator Ops Finance and Business API and Retail Customers Integrations & Projects 12. Scaling for Customers 13. In the beginning 14. Single Instance 15. Web Instances Single DB Instance Elastic Load Balancer 16. Going Global 17. Web Instances Single DB Instance Elastic Load Balancer CloudFront 18. Where are we going to keep it all? Enter S3. 19. Web Instances Single DB Instance Elastic Load Balancer CloudFront S3 20. Web Instances Single DB Instance Elastic Load Balancer CloudFront S3 Redis Instances 21. Asynchronous Jobs Processing 22. Web Instances Single DB Instance Elastic Load Balancer CloudFront S3 Redis Instances Processor Instances 23. Scaling for Translators 24. More space! 25. Growth - 10,000,000 20,000,000 30,000,000 40,000,000 50,000,000 60,000,000 70,000,000 80,000,000 90,000,000 100,000,000 2008 2009 2010 2011 2012 2013 Units Units Translated 26. Complex Workflows 27. 100% CPU 28. Web Instances Single R/W DB Instance Elastic Load Balancer CloudFront S3 Redis Instances Processor Instances Multiple Read DB Instance 29. Read Only Slaves 30. Web Instances Elastic Load Balancer S3 Redis Instances Processor InstancesDB Instances API Instances Elastic Load Balancer Segregating concerns 31. Caching! 32. Scaling for Operations 33. Growing up 34. Dedicated Ops DB Slave 35. Translator Operations 36. Search for Operations 37. Breaking it out further 38. Web Instances Elastic Load Balancer S3 Redis Instances Processor InstancesDB Instances API Instances Elastic Load Balancer Admin Instances Elastic Load Balancer Retail Instances Elastic Load Balancer 39. Lessons Learned + Future 40. Web Instances Elastic Load Balancer S3 ElastiCache Processor InstancesRDS DB Instances API Instances Elastic Load Balancer Admin Instances Elastic Load Balancer Retail Instances Elastic Load Balancer 41. Measure Your Application not just your servers! 42. Give Your Applications a Voice 43. Monitor Everything 44. Wow Such thanks.