Qconsp domesticando dragoes com solu??es escal¨¢veis
?
2 likes?352 views
The document discusses the challenges of building a scalable SMS notification system. An initial pilot system could only handle 3 transactions per second, but production needs were over 17 transactions per second with peaks of 200 per second. To solve this, the system was improved by using message queues and a Redis cache to handle the higher volumes and meet the client's needs.
Qconsp domesticando dragoes com solu??es escal¨¢veis
- 1. Domesticando Drag?es com solu??es escal¨¢veis @abstractj @qmx Monday, September 13, 2010
- 2. O cliente pediu um sistema Monday, September 13, 2010
- 3. Envio de SMS pela intranet + aplicacao web pra ver detalhes da mensagem Monday, September 13, 2010
- 4. A cl¨¢usula ¡°O sistema deve fazer tudo¡± Caso de uso, 2004 Monday, September 13, 2010
- 5. Teu chefe Monday, September 13, 2010
- 6. F¨¢cil! Monday, September 13, 2010
- 7. Neg¨®cio fechado Monday, September 13, 2010
- 8. SEU EMO!!!! Monday, September 13, 2010
- 9. Sistema em piloto 10.000 tx/h = quase 3 tx/s Monday, September 13, 2010
- 10. Bonito e Perform¨¢tico Monday, September 13, 2010
- 11. Cliente satisfeito Monday, September 13, 2010
- 12. A volta dos que n?o foram Monday, September 13, 2010
- 13. A cl¨¢usula ¡°O sistema deve fazer tudo¡± Caso de uso, 2004 Monday, September 13, 2010
- 15. O legado Monday, September 13, 2010
- 16. Moleza! Monday, September 13, 2010
- 17. Legado = Mainframe Monday, September 13, 2010
- 18. Ouch! Monday, September 13, 2010
- 19. Mainframe Monday, September 13, 2010
- 20. N?o gosta dos seus dados Monday, September 13, 2010
- 21. N?o quer trabalhar pra voc¨º Monday, September 13, 2010
- 22. N?o espera, voc¨º espera Monday, September 13, 2010
- 23. te odeia Monday, September 13, 2010
- 24. Mainframe n?o escala Monday, September 13, 2010
- 25. Mainframe n?o escala? Monday, September 13, 2010
- 26. Motivos t¨¦cnicos Monday, September 13, 2010
- 27. FUUUUUU Monday, September 13, 2010
- 28. #medo? Monday, September 13, 2010
- 29. Ainda bem Monday, September 13, 2010
- 30. Sobrou! Monday, September 13, 2010
- 31. Flat ?les? Monday, September 13, 2010
- 32. batchizando? Monday, September 13, 2010
- 33. Filas de Mensageria! Monday, September 13, 2010
- 34. Problema resolvido! 3 tx/s Monday, September 13, 2010
- 35. A cl¨¢usula ¡°O sistema deve fazer tudo¡± Caso de uso, 2004 Monday, September 13, 2010
- 36. Projeto piloto 3 tx/s = Bom Monday, September 13, 2010
- 37. Realidade: 1.5M tx/dia = 17tx/s Monday, September 13, 2010
- 38. Estamos devendo: ¡À 1.25M tx Monday, September 13, 2010
- 39. "Compra de R$ 5000,00 realizada com sucesso em 10/09/2010 03:56" 12/09/2010 15:30 Monday, September 13, 2010
- 40. Murphy: se voc¨º corria risco de fraude, ela aconteceu com certeza Monday, September 13, 2010
- 41. Realidade? Monday, September 13, 2010
- 42. Pico de 200tx/s na hora do almo?o Monday, September 13, 2010
- 43. E agora? Monday, September 13, 2010
- 44. Onde foi que eu errei? Monday, September 13, 2010
- 45. Voc¨º ¨¦ culpado Monday, September 13, 2010
- 46. Na m¨¦dia, o ser humano tem uma bola Monday, September 13, 2010
- 47. Voc¨º perguntou pro seu cliente quais eram os hor¨¢rios de pico?! Monday, September 13, 2010
- 48. Sua obriga??o era medir!!! Monday, September 13, 2010
- 49. Sua aplica??o escala?! Monday, September 13, 2010
- 50. Voc¨º vai at¨¦ onde na vertical?! Monday, September 13, 2010
- 51. E na horizontal?! Monday, September 13, 2010
- 52. Voc¨º n?o est¨¢ esquecendo de nada? Monday, September 13, 2010
- 53. Visualizar os detalhes da mensagem Monday, September 13, 2010
- 54. Ruby ¨¦ lento? Monday, September 13, 2010
- 55. A culpa ¨¦ de quem? Monday, September 13, 2010
- 56. Banco de dados? Monday, September 13, 2010
- 57. Massa de dados grande Monday, September 13, 2010
- 58. 1.500.000 x 30 dias Monday, September 13, 2010
- 59. Em um m¨ºs 45 milh?es de registros Monday, September 13, 2010
- 60. preciso guardar tudo isso? Monday, September 13, 2010
- 61. relacional pra qu¨º? Monday, September 13, 2010
- 62. Cache? Monday, September 13, 2010
- 63. #fail Monday, September 13, 2010
- 64. Cache persistente? Monday, September 13, 2010
- 65. Redis key-value store on steroids Monday, September 13, 2010
- 66. O melhor dos dois mundos Monday, September 13, 2010
- 67. Dados bem guardados com SQL e com NoSQL Monday, September 13, 2010
- 68. Pesquisa direta no ¡°banco de dados¡± 20tx/s Monday, September 13, 2010
- 69. Pesquisa passando pelo Redis: 200 tx/s Monday, September 13, 2010
- 70. Seus problemas acabaram! Monday, September 13, 2010
- 71. Obrigado! Monday, September 13, 2010